STARD 2015: an updated list of essential items for reporting diagnostic accuracy studies

Visualized hysteroscopic artificial intelligence fertility assessment system for endometrial injury: an image-deep-learning study

OBJECTIVE

Asherman's syndrome (AS) is a significant cause of subfertility in women from developing countries. Over 80% of AS cases in these regions are linked to dilation and curettage (D&C) procedures following pregnancy. The incidence of AS in patients with infertility and recurrent miscarriage can be as high as 10%, while the pregnancy rate in cases of moderate to severe adhesions can be as low as 34%. We aimed to establish a hysteroscopic artificial intelligence system using image-deep-learning algorithms for fertility assessment.

METHODS

This diagnostic study included 555 cases with 4922 hysteroscopic images from a Chinese intrauterine adhesions cohort clinical database (NCT05381376). The study evaluated two image-deep-learning algorithms' effectiveness in predicting pregnancy within one year, using AUCs and decision curve analysis. The models' performance was evaluated for two-year prediction via concordance index and cumulative time-dependent ROC. A quantifiable visualization panel of the system was established.

RESULTS

The proportional hazard CNN system accurately predicted conception, with AUCs of 0.982, 0.992, and 0.990 in three randomly assigned datasets, superior to the InceptionV3 framework, and achieved a net benefit of 69.4% for subfertility assessment. The system fitted well with c-indexes of 0.920-0.940 and was time-stable. The quantifiable visualization panel displayed four intrauterine pathologies intuitively. The performance was comparable to senior hysteroscopists, with a kappa value of 0.84-0.89.

CONCLUSIONS

The CNN based on the proportional hazard approach accurately assesses fertility postoperatively. The quantifiable visualization panel could assist in intrauterine pathologies assessment, optimize treatment strategies, and achieve individualized and cost-efficient practices.

Underestimation of Follow-Up Infarct Volume by Acute CT Perfusion Imaging

BACKGROUND AND OBJECTIVES

It is unknown whether acute CT perfusion (CTP) core imaging may underestimate the follow-up infarct. We hypothesize that infarct underestimation occurs especially in late-presenting patients and that underestimated infarct can partially be detected on baseline noncontrast CT (NCCT).

METHODS

We included patients with acute anterior circulation ischemic stroke who underwent baseline NCCT and CTP imaging, complete endovascular reperfusion, and follow-up MRI from the Endovascular Therapy Following Imaging Evaluation for Ischemic Stroke (DEFUSE 3) trial and a consecutive, monocenter cohort. We divided patients into early (<6 hours) and late (6-24 hours) presenters. We performed semiautomated segmentations of the acute ischemic lesion on NCCT using 5% relative density difference (rNCCT>5%) and used the relative cerebral blood flow <30% to segment the CTP core. On coregistered images, we performed volumetric and voxel-based analyses to compare infarct estimations by imaging modality. Spatial accuracy for the follow-up infarct was assessed using the Dice similarity coefficient (DSC) and balanced accuracy.

RESULTS

We included 109 patients with a median age of 70 (interquartile range [IQR] 31-93) years of whom 52% were female. The follow-up infarct was underestimated by the CTP core (mean absolute volume difference [MAVD] = 14 mL [SD 36], p < 0.001), but not by the union lesion (MAVD = 3 mL [SD 32], p = 0.76). Infarct underestimation was greater in late presenters (median 17 mL [IQR 7-33] vs 7 mL [IQR 4-25] in early presenters, p < 0.01) and in patients with poor collaterals (median 20 mL [IQR 8-56] vs 8 mL [IQR 4-20] in patients with good collaterals, p < 0.01). Median 25% of the infarct missed by the CTP core could be detected on baseline rNCCT in late presenters (vs. median 3% in early presenters). The combined rNCCT>5% and CTP core lesion more accurately detected the follow-up infarct compared with the CTP core alone (median DSC 0.37 [IQR 0.06-0.55] vs 0.18 [IQR 0-0.42] and median balanced accuracy 0.67 [IQR 0.53-0.75] vs 0.56 [IQR 0.50-0.67], p < 0.001 for both).

DISCUSSION

Underestimation of follow-up infarct by CTP is substantial and the follow-up infarct can partially be detected by baseline NCCT, especially in patients with stroke with delayed presentation. Combining rNCCT>5% and CTP increases the accuracy for predicting the follow-up infarct.

Subtyping strokes using blood-based protein biomarkers: A high-throughput proteomics and machine learning approach

BACKGROUND

Rapid diagnosis of stroke and its subtypes is critical in early stages. We aimed to discover and validate blood-based protein biomarkers to differentiate ischemic stroke (IS) from intracerebral haemorrhage (ICH) using high-throughput proteomics.

METHODS

We collected serum samples within 24 h from acute stroke (IS & ICH) and mimics patients. In the discovery phase, SWATH-MS proteomics identified differentially expressed proteins, which were validated using targeted proteomics in the validation phase. We conducted interaction network and pathway analyses using Cytoscape 3.10.0. We determined cut-off points using the Youden Index. We developed three prediction models using multivariable logistic regression analyses. We assessed the model performance using statistical tests.

RESULTS

We included 20 IS and 20 ICH in the discovery phase and 150 IS, 150 ICH, and six stroke mimics in the validation phase. We quantified 375 proteins using SWATH-MS. Between IS and ICH, we discovered 20 differentially expressed proteins. In the validation phase, the combined prediction model including three biomarkers: GFAP (aOR 0.04; 95%CI .02-.11), MMP-9 (aOR .09; .03-.28), APO-C1 (aOR 5.76; 2.66-12.47) and clinical variables independently differentiated IS from ICH (accuracy: 92%, negative predictive value: 94%). Adding biomarkers to clinical variables improved discrimination by 26% (p < .001). Additionally, nine biomarkers differentiated IS from ICH within 6 h, while three biomarkers differentiated IS from mimics.

CONCLUSIONS

Our study demonstrated that GFAP, MMP-9 and APO-C1 biomarkers independently differentiated IS from ICH within 24 h and significantly improved the discrimination ability of prediction models. Temporal profiling of these biomarkers in the acute phase of stroke is warranted.

Development and validation of a prediction model for malignant sinonasal tumors based on MR radiomics and machine learning

OBJECTIVES

This study aimed to utilize MR radiomics-based machine learning classifiers on a large-sample, multicenter dataset to develop an optimal model for predicting malignant sinonasal tumors and tumor-like lesions.

METHODS

This study included 1711 adult patients (875 benign and 836 malignant) with sinonasal tumors or tumor-like lesions from three institutions. Patients from institution 1 (n = 1367) constituted both the training and validation cohorts, while those from institution 2 and 3 (n = 158/186) made up the test cohorts. Manual segmentation of the region of interest of the tumor was performed on T1WI, T2WI, and contrast-enhanced T1WI (CE-T1WI). Data normalization, dimensional reductions, feature selection, and classifications were performed using ten machine-learning classifiers. Four fusion models, namely T1WI + T2WI, T1WI + CE-T1WI, T2WI + CE-T1WI, and T1WI + T2WI + CE-T1WI, were constructed using the top ten features with the highest contribution in feature selection in the optimal models of T1WI, T2WI, and CE-T1WI. The Delong test compared areas under the curve (AUC) between models.

RESULTS

The AUCs of training/validation/test1/test2 datasets for T1WI, T2WI, and CE-T1WI were 0.900/0.842/0.872/0.839, 0.876/0.789/0.842/0.863, and 0.899/0.824/0.831/0.707, respectively. The fusion model from T1WI + T2WI + CE-T1WI had the highest AUC. The AUCs of training/validation/test1/test2 datasets were 0.947/0.849/0.871/0.887. The T1WI + T2WI + CE-T1WI model demonstrated a significantly higher AUC than the T2WI + CE-T1WI model in both cohorts (p < 0.05) and outperformed the T2WI model in test 1 (p = 0.008) and the T1WI model in test 2 (p = 0.006).

CONCLUSIONS

This fusion model based on radiomics from T1WI + T2WI + CE-T1WI images and machine learning can improve the power in predicting malignant sinonasal tumors with high accuracy, resilience, and robustness.

CLINICAL RELEVANCE STATEMENT

Our study proposes a radiomics-based machine learning fusion model from T1- and T2-weighted images and contrast-enhanced T1-weighted images, which can non-invasively identify the nature of sinonasal tumors and improve the performance in predicting malignant sinonasal tumors.

KEY POINTS

Differentiating benign and malignant sinonasal tumors is difficult due to similar clinical presentations. A radiomics model from T1 + T2 + contrast-enhanced T1 images can identify the nature of sinonasal tumors. This model can help distinguish benign and malignant sinonasal tumors.

Diagnostic Algorithms for Adnexal Masses in the Hands of a Novice Operator

OBJECTIVE

To compare the performance of four commonly used algorithms to differentiate benign from malignant adnexal masses when used by a novice operator.

METHODS

Women with adnexal masses treated at Mayo Clinic, Rochester, Minnesota, in 2019 were identified retrospectively. Patients were included if they underwent surgery within 3 months of diagnosis or had at least 10 months of follow-up. A nonexpert operator (European Federation of Societies for Ultrasound in Medicine and Biology level I) classified each lesion using ADNEX (Assessment of Different Neoplasias in the Adnexa), two-step strategy (benign descriptors followed by ADNEX), O-RADS (Ovarian-Adnexal Reporting and Data System) 2019, and O-RADS 2022. The primary outcome measure was the area under the receiver operating characteristic curve (AUC) compared across the four algorithms.

RESULTS

A total of 556 women were included in the analyses: 452 with benign and 104 with malignant masses. The AUCs of ADNEX, the two-step strategy, O-RADS 2019, and O-RADS 2022 were 0.90 (95% CI, 0.87-0.94), 0.91 (95% CI,0.88-0.94), 0.88 (95% CI,0.84-0.91), and 0.88 95% CI, (0.84-0.91), respectively. The two-step strategy performed significantly better than the O-RADS algorithms ( P =.005 and P =.002). With all the algorithms, the observed malignancy rate was 1.9-2.2% among lesions categorized as almost certainly benign, twofold higher than the expected less than 1.0%. Lesions wrongly classified as almost certainly benign were borderline tumors (n=4) and metastases (n=3).

CONCLUSION

In the hands of a novice operator, all algorithms performed well and were able to distinguish benign from malignant lesions. Although the two-step strategy performed slightly better than the O-RADSs, the difference did not appear to be clinically meaningful. The malignancy rate among lesions classified as almost certainly benign was unexpectedly high at 1.9-2.3%, approximately double the expected rate of less than 1.0%.

Diagnostic accuracy of tongue coating in identifying acute appendicitis: a prospective cohort study

BACKGROUND

Acute appendicitis requires timely diagnosis. The diagnostic efficacy of tongue examination in making this diagnosis has not been established. This study investigates whether the Tongue Coating Index (TCI), a validated measure of tongue coating, can aid in diagnosing acute appendicitis.

METHODS

We conducted a prospective cohort study (1 September 2018-31 December 2020) at a single Japanese hospital. Adults (≥20 years) with suspected acute appendicitis, presenting to either the emergency department or general outpatient clinic, were enrolled. Tongue images were taken at presentation; two independent examiners-unrelated to clinical care and blinded to patient data-later scored these images using the TCI. A composite reference standard (clinical findings, imaging, histopathology, follow-up) was used to confirm appendicitis. We compared the TCI's diagnostic performance with the Alvarado score and its components using C-index, area under the curve (AUC), sensitivity and specificity.

RESULTS

Of 145 included patients, 69 (47.6%) were diagnosed with acute appendicitis. The TCI demonstrated comparable discriminative ability (C-index AUC 0.62; 95% CI, 0.53 to 0.71) to that of the Alvarado score (0.66; 95% CI, 0.57 to 0.75). Of Alvarado score components, migration of pain had an AUC of 0.63 (95% CI, 0.55 to 0.71), anorexia 0.58 (95% CI, 0.50 to 0.66) and tenderness in the right lower quadrant 0.55 (95% CI, 0.50 to 0.60). At a cut-off of 3, the TCI demonstrated high sensitivity of 96% (95% CI, 88% to 98%) but low specificity of 21% (95% CI, 13% to 32%). Conversely, at a cut-off of 10, the TCI showed increased specificity of 83% (95% CI, 73% to 90%) but reduced sensitivity of 29% (95% CI, 20% to 41%).

CONCLUSION

The TCI showed comparable diagnostic performance to the Alvarado score and its individual components. TCI may potentially serve as an additional non-invasive indicator for diagnosing or ruling out acute appendicitis. Further research is essential to validate its clinical utility.

Comparison of Ultrasonography, CBCT, Transgingival Probing, Colour-Coded and Periodontal Probe Transparency With Histological Gingival Thickness: A Diagnostic Accuracy Study Revisiting Thick Versus Thin Gingiva

AIM

To assess the reliability of ultrasonographic, cone beam computed tomographic (CBCT), probe transparency and transgingival probing (TGP) methods in evaluating gingival thickness (GT), compared with the gold standard histological assessment.

METHODS

Sixteen fresh cadaver heads with intact gingivae were used. The sequence for GT measurement included CBCT, ultrasonography, probe transparency, TGP and histology. Both stainless steel periodontal probe and colour-coded probes were used for transparency. TGP involved a calibrated endodontic spreader, and histologic samples served as a comparative standard. Primary outcomes evaluated accuracy in GT measurement, while secondary outcomes assessed agreement among methods and established an optimal threshold for thin versus thick gingiva.

RESULTS

One hundred and fifteen teeth were examined, yielding a mean GT of 1.34 mm histologically. US and CBCT underestimated GT (means of 1.25 mm and 1.13 mm, respectively), while TGP overestimated (1.51 mm). Correlations (r = 0.88-0.98) and ICC values (0.73-0.95) indicated strong inter-method agreement. Regression models significantly estimated histological GT from US, CBCT and TGP. A new 1.18 mm cut-off, based on histology, improved diagnostic accuracy over the traditional 1 mm threshold.

CONCLUSIONS

While histology remains the GT reference standard, US, CBCT and TGP achieved clinically acceptable accuracy. US showed the highest agreement with histology, followed by TGP and CBCT. The study supports US as the most practical non-invasive tool, although CBCT and TGP remain viable options. Further clinical validation is recommended, acknowledging the limitations of cadaveric models in reflecting in vivo conditions.

Accuracy and Reliability of Intraoral 3D Scans for Diagnostic Evaluations in Nursing Home Residents

OBJECTIVES

To evaluate the accuracy and reliability of intraoral 3D scans for assessing the oral health of older patients living in nursing homes.

MATERIALS AND METHODS

One examiner recorded missing teeth, restorations, caries lesions and oral hygiene (Geriatric-Debris-Index Simplified [GDI-S] ≥ 1.9) in nursing home residents using visual-tactile diagnostics (reference test) and afterwards obtained intraoral scans with an intraoral scanner (TRIOS 4, 3Shape). Two other independent investigators and the clinical examiner assessed only the scans to delineate the same diagnostic outcomes. For these outcomes, we assessed accuracy, sensitivity, specificity and the Area-under-the-Receiver-Operating-Characteristics Curve (AUROC). Kappa values (κ) were calculated to evaluate inter-examiner and intra-examiner reliability after re-examination by all examiners after a minimum interval of 3 weeks.

RESULTS

Forty-three partially dentate patients (65-95 years of age) in need of care living in four nursing homes with a total of 486 teeth (mean [SD]: 11 [9] teeth per patient) were examined. Scans were perfectly accurate for detecting missing teeth (AUROC [sensitivity/specificity]: 1 [1/1]) and showed high accuracy for detecting restorations (0.96 [0.93/0.98]), too. Accuracy was lower to detect caries (0.77 [0.58/0.97]) and insufficient for oral hygiene (0.76 [0.54/0.99]). Agreement between examiners was perfect for missing teeth (κ: 1), very good for the detection of restorations (0.94), and good for caries or insufficient oral hygiene (0.73 and 0.72, respectively).

CONCLUSIONS

Scans were suitable for basic diagnostic evaluations but showed considerable shortcomings in detecting caries and poor oral hygiene. Assessing scans was relatively reliable.

CLINICAL RELEVANCE

Using scans may allow telemedical assessments of nursing home residents, but users should be aware of the differential accuracy for different diagnostic targets.

Artificial intelligence for intraoperative video analysis in robotic-assisted esophagectomy

BACKGROUND

Robotic-assisted minimally invasive esophagectomy (RAMIE) is a complex surgical procedure for treating esophageal cancer. Artificial intelligence (AI) is an uprising technology with increasing applications in the surgical field. This scoping review aimed to assess the current AI applications in RAMIE, with a focus on intraoperative video analysis.

METHODS

To identify all articles utilizing AI in RAMIE, a comprehensive literature search was performed in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-analysis for scoping reviews of Medline and Embase databases and the Cochrane Library. Two independent reviewers assessed articles for quality and inclusion.

RESULTS

One hundred and seventeen articles were identified, of which four were included in the final analysis. Results demonstrated that the main AI applications in RAMIE were intraoperative video assessment and the evaluation of surgical technical skills to evaluate surgical performance. AI was also used for surgical phase recognition to support clinical decision-making through intraoperative guidance and identify key anatomical landmarks. Various deep-learning networks were used to generate AI models, and there was a strong emphasis on using high-quality standardized video frames.

CONCLUSIONS

The use of AI in RAMIE, especially in intraoperative video analysis and surgical phase recognition, is still a relatively new field that should be further explored. The advantages of using AI algorithms to evaluate intraoperative videos in an automated manner may be harnessed to improve technical performance and intraoperative decision-making, achieve a higher quality of surgery, and improve postoperative outcomes.

Diagnostic efficacy of an extracellular vesicle-derived lncRNA-based liquid biopsy signature for the early detection of early-onset gastric cancer

BACKGROUND

Early-onset gastric cancer (EOGC) is a lethal malignancy. It differs from late-onset gastric cancer (LOGC) in clinical and molecular characteristics. The current strategies for EOGC detection have certain limitations in diagnostic performance due to the rising trend in EOGC.

OBJECTIVE

We developed a liquid biopsy signature for EOGC detection.

DESIGN

We use a systematic discovery approach by analysing genome-wide transcriptomic profiling data from EOGC (n=43), LOGC (n=31) and age-matched non-disease controls (n=37) tissue samples. An extracellular vesicle-derived long non-coding RNA (EV-lncRNA) signature was identified in blood samples from a training cohort (n=299), and subsequently confirmed by qPCR in two external validation cohorts (n=462 and n=438), a preoperative/postoperative cohort (n=66) and a gastrointestinal tumour cohort (n=225).

RESULTS

A three EV-lncRNA (NALT1, PTENP1 and HOTTIP) liquid biopsy signature was developed for EOGC detection with an area under the receiver operating characteristic curve (AUROC) of 0.924 (95% CI 0.889 to 0.953). This EV-lncRNA signature provided robust diagnostic performance in two external validation cohorts (Xi'an cohort: AUROC, 0.911; Beijing cohort: AUROC, 0.9323). Furthermore, the EV-lncRNA signature reliably identified resectable stage EOGC patients (stage I/II) and demonstrated better diagnostic performance than traditional GC-related biomarkers in distinguishing early-stage EOGC (stage I) from precancerous lesions. The low levels of this biomarker in postsurgery and other gastrointestinal tumour plasma samples indicated its GC specificity.

CONCLUSIONS

The newly developed EV-lncRNA signature effectively identified EOGC patients at a resectable stage with enhanced precision, thereby improving the prognosis of patients who would have otherwise missed the curative treatment window.

The validation of the CTS-6 Evaluation Tool for diagnosing carpal tunnel syndrome (CTS) in Thai wheelchair users

OBJECTIVE

The CTS-6 Evaluation Tool is a clinical diagnostic tool for carpal tunnel syndrome. It was originally developed using data from non-disabled individuals and has never been validated in different population. This study aimed to validate CTS-6's diagnostic performance at a cutoff score of 12 in a new population of wheelchair users.

METHODS

The participants were 54 Thai wheelchair users from a university hospital's neuropathy registry. Those with a history of nerve injury, fracture, neuropathy, or pregnancy were excluded from the study. All underwent clinical exam, CTS-6, and electrodiagnosis (blinded evaluator). Carpal tunnel syndrome was diagnosed based on clinical symptoms and electrodiagnostic criteria.

RESULTS

Of 54 participants, 13 were female (24.1%) with an average age of 46.9 (SD 12.2) years, and 18 (33.3%) participants had carpal tunnel syndrome. Duration of disability (years) was significantly longer in cases; median 24 (Q1 19.0, Q3 28.9), p < 0.001. Discriminative performance: Area under the receiver operating curve 0.935 (95%CI:0.891-0.978). At the 12 cutoff point, the sensitivity was 43.8% (95%CI:26.4-62.3%) and the specificity was 100.0% (95%CI:94.8-100.0%). A lower cutoff point showed increased sensitivity and specificity. Symptomatic subgroup analysis showed similar diagnostic performances.

CONCLUSION

The CTS-6 Evaluation Tool is a simple clinical diagnostic tool that does not require sophisticated investigation. The CTS-6's discriminative ability remains strong. The diagnostic performance at a cutoff score of 12 showed moderate sensitivity and high specificity. Applying a cutoff score of 12 could help rule in the diagnosis where access to electrodiagnosis is limited. A lower cutoff score that is 7.5 could be applied as a screening test to rule out the diagnosis, as it provides moderately higher sensitivity at the cost of increased false positives.

Generalized Anxiety Disorder 7-item (GAD-7) and 2-item (GAD-2) scales for detecting anxiety disorders in adults

BACKGROUND

Anxiety disorders often remain undetected and can cause substantial burden. Amongst the many anxiety screening tools, the 7-item Generalized Anxiety Disorder (GAD-7) scale and its short version, the 2-item Generalized Anxiety Disorder (GAD-2) scale, are the most frequently used instruments.

OBJECTIVES

Primary: to determine the diagnostic accuracy of GAD-7 and GAD-2 to detect generalised anxiety disorder (GAD) and any anxiety disorder (AAD) in adults. Secondary: to investigate whether their diagnostic accuracy varies by setting, anxiety disorder prevalence, reference standard, and risk of bias; to compare the diagnostic accuracy of GAD-7 and GAD-2; to investigate how diagnostic performance changes with the test threshold.

SEARCH METHODS

We searched MEDLINE, Embase, PubMed-not-MEDLINE subset, and PsycINFO from 1990 to 18 January 2024. We checked reference lists of included studies and review articles.

SELECTION CRITERIA

We included cross-sectional studies conducted in adults, containing diagnostic accuracy information on GAD-7 and/or GAD-2 questionnaires for the target conditions generalised anxiety disorder and/or any anxiety disorder, and allowing the generation of 2x2 tables. The target conditions must have been diagnosed using a structured or semi-structured clinical interview. We excluded case-control studies and studies in which the time elapsed between the index tests and reference standards exceeded four weeks. We excluded studies involving people (1) seeking help in mental health settings or (2) recruited specifically due to mental health symptoms in other settings.

DATA COLLECTION AND ANALYSIS

At least two review authors independently decided on study eligibility, extracted data, and assessed the risk of bias and applicability of included studies. For each questionnaire and each target condition, we present sensitivity and specificity with 95% confidence intervals (95% CI) in forest plots. We used the bivariate model to obtain summary estimates based on cut-offs closest to the recommended values (i.e. within a core range). In secondary analyses, we used the bivariate model and the multiple thresholds model to obtain summary estimates for all available cut-off points. Using the multiple thresholds model, we also calculated the area under the receiver operating characteristic curve to obtain a general indicator of the diagnostic accuracy of GAD-7 and GAD-2.

MAIN RESULTS

We included 48 studies with 19,228 participants from 27 different countries, evaluating the GAD-7 and the GAD-2 in 24 different languages. Seven studies were performed in non-clinical settings, nine in clinical settings recruiting participants across conditions, and 32 in clinical settings with participants having specific conditions. Even after categorisation into three settings, the study populations were substantially different. The most frequently studied populations were people: with epilepsy (nine studies); with cancer (five studies); with cardiovascular disease (five studies); and in primary care regardless of their condition (five studies). We considered the risk of bias low in eight studies, and we had low concerns about the applicability of findings in three studies. Thirty-five studies contributed to the primary analyses of GAD-7 for detecting generalised anxiety disorder (median prevalence 12%); 22 studies to analyses of GAD-7 for any anxiety disorder (median prevalence 19%); 24 studies to analyses of GAD-2 for generalised anxiety disorder (median prevalence 9%); and 19 studies to analyses of GAD-2 for any anxiety disorder (median prevalence 19%). At the recommended cut-off of 10 or higher (or the closest available cut-off), the GAD-7 questionnaire yielded a summary sensitivity of 0.64 (95% CI 0.56 to 0.72) and a summary specificity of 0.91 (95% CI 0.87 to 0.93) in detecting generalised anxiety disorder. For detecting any anxiety disorder, summary sensitivity was 0.48 (95% CI 0.40 to 0.57) and summary specificity 0.91 (95% CI 0.89 to 0.93). At the recommended cut-off of 3 or higher (or the closest available cut-off), the GAD-2 yielded a summary sensitivity of 0.68 (95% CI 0.59 to 0.75) and a summary specificity of 0.86 (95% CI 0.82 to 0.89) for detecting generalised anxiety disorder. For detecting any anxiety disorder, the summary sensitivity was 0.53 (95% CI 0.44 to 0.62) and the summary specificity was 0.89 (95% CI 0.86 to 0.91). The 95% prediction region of GAD-7 for detecting generalised anxiety disorder was larger (indicating pronounced statistical heterogeneity) than for the three other analyses. Specificity varied by setting in the analysis of GAD-7 and GAD-2 for detecting any anxiety disorder, and by reference standard in the analysis of GAD-2 for detecting generalised anxiety disorder. Sensitivity varied with prevalence in the analysis of GAD-7 for generalised anxiety disorder. Other investigations of potential sources of heterogeneity did not show statistically significant associations with test accuracy. In all analyses, sensitivity tended to be higher and specificity lower in participants with specific conditions compared to the other two settings. Overall, the heterogeneity in the subgroup analyses remained high. The area under the receiver operating characteristic curve in the multiple thresholds model was 0.86 (95% CI 0.84 to 0.88) for the GAD-7 scale in detecting generalised anxiety disorder, and 0.80 (95% CI 0.78 to 0.82) in detecting any anxiety disorders. For the GAD-2 scale, the value was 0.82 (95% CI 0.81 to 0.86) for detecting generalised anxiety disorder, and 0.77 (95% CI 0.76 to 0.82) for detecting any anxiety disorders. Comparative bivariate analyses revealed no statistically significant differences between the diagnostic test accuracy of GAD-7 and GAD-2.

AUTHORS' CONCLUSIONS

The GAD-7 and the GAD-2 scales have been tested in numerous languages and different populations. Overall, the GAD-7 and the GAD-2 seem to have acceptable or good diagnostic accuracy for both generalised anxiety disorder and any anxiety disorder. The GAD-2 scale seems to have similar diagnostic accuracy as the GAD-7 scale. However, due to the diversity of the included studies and the heterogeneity of our findings, our summary estimates of sensitivity and specificity should be interpreted as rough averages. The performance of GAD-7 and GAD-2 may deviate substantially from these values in specific situations.

Annotation-efficient, patch-based, explainable deep learning using curriculum method for breast cancer detection in screening mammography

OBJECTIVES

To develop an efficient deep learning (DL) model for breast cancer detection in mammograms, utilizing both weak (image-level) and strong (bounding boxes) annotations and providing explainable artificial intelligence (XAI) with gradient-weighted class activation mapping (Grad-CAM), assessed by the ground truth overlap ratio.

METHODS

Three radiologists annotated a balanced dataset of 1976 mammograms (cancer-positive and -negative) from three centers. We developed a patch-based DL model using curriculum learning, progressively increasing patch sizes during training. The model was trained under varying levels of strong supervision (0%, 20%, 40%, and 100% of the dataset), resulting in baseline, curriculum 20, curriculum 40, and curriculum 100 models. Training for each model was repeated ten times, with results presented as mean ± standard deviation. Model performance was also tested on an external dataset of 4276 mammograms to assess generalizability.

RESULTS

F1 scores for the baseline, curriculum 20, curriculum 40, and curriculum 100 models were 80.55 ± 0.88, 82.41 ± 0.47, 83.03 ± 0.31, and 83.95 ± 0.55, respectively, with ground truth overlap ratios of 60.26 ± 1.91, 62.13 ± 1.2, 62.26 ± 1.52, and 64.18 ± 1.37. In the external dataset, F1 scores were 74.65 ± 1.35, 77.77 ± 0.73, 78.23 ± 1.78, and 78.73 ± 1.25, respectively, maintaining a similar performance trend.

CONCLUSION

Training DL models with a curriculum method and a patch-based approach yields satisfactory performance and XAI, even with a limited set of densely annotated data, offering a promising avenue for deploying DL in large-scale mammography datasets.

CRITICAL RELEVANCE

This study introduces a DL model for mammography-based breast cancer detection, utilizing curriculum learning with limited, strongly labeled data. It showcases performance gains and better explainability, addressing challenges of extensive dataset needs and DL's "black-box" nature.

KEY POINTS

Increasing numbers of mammograms for radiologists to interpret pose a logistical challenge. We trained a DL model leveraging curriculum learning with mixed annotations for mammography. The DL model outperformed the baseline model with image-level annotations using only 20% of the strong labels. The study addresses the challenge of requiring extensive datasets and strong supervision for DL efficacy. The model demonstrated improved explainability through Grad-CAM, verified by a higher ground truth overlap ratio. He proposed approach also yielded robust performance on external testing data.

Closing the gap in kidney disease: validating the reporting of Aboriginal and/or Torres Strait Islander identification in a clinical quality registry using linked data

OBJECTIVE

To examine the accuracy of the Australia and New Zealand Dialysis and Transplant Registry (ANZDATA), the population-based clinical quality registry for people with kidney failure, in identifying Aboriginal and/or Torres Strait Islander people.

DESIGN

Population-based cohort study of reporting accuracy.

SETTING

New South Wales, 2006-2020.

PARTICIPANTS

Incident kidney failure patients.

MAIN OUTCOME MEASURES

Sensitivity and specificity of identification of Aboriginal and/or Torres Strait Islander people in ANZDATA compared with identification with Enhanced Reporting of Aboriginality (ERA) methods using linked health datasets.

RESULTS

Of 11 708 patients, 693 (5.9%) were identified as Aboriginal and/or Torres Strait Islander people using ERA methods, with 484 recognised in ANZDATA. Overall ANZDATA sensitivity was 67.0% (95% CI, 63.3-70.5%), with high specificity (99.8%; 95% CI, 99.7-99.9%). Sensitivity was lowest for males (63.8%; 95% CI, 58.7-68.6), people aged under 18 years (45.0%; 95% CI, 23.1-68.5%) or over 65 years (61.7%; 95% CI, 53.8-69.2%), and those with greater socio-economic advantage (56.6%; 95% CI, 46.6-66.2%), living in major cities (53.8%; 95% CI, 48.0-59.5%) and with no comorbidities (47.7%; 95% CI, 37.0-58.6%). Aboriginal and/or Torres Strait Islander people identified in ANZDATA had lower rates of waitlisting for kidney transplantation (17.8% v 25.3%; P = 0.016) and receiving a kidney transplant (12.2% v 23.1%; P < 0.001) and a higher rate of death (56.0% v 44.5%; P = 0.004) compared with those not recognised in ANZDATA.

CONCLUSION

Aboriginal and/or Torres Strait Islander people were under-reported in ANZDATA. There were multiple biases in characteristics and outcomes for people identified in ANZDATA compared with those identified by ERA using linked data. This highlights the importance of data integration as a quality improvement mechanism and identifying barriers to disclosure.

Enhancing precision in sarcoma diagnosis: nCounter fusion panel implementation in a middle-income country

Background

Sarcoma diagnosis is challenging due to numerous subtypes with similar histopathological features and the high cost of fusion detection methods, particularly in middle-income countries.

Objectives

To implement a cost-effective custom-based nCounter approach previously validated for fusion analysis of suspected sarcoma in Brazil.

Design and methods

RNA isolated from 56 routine sarcomas, which were formalin-fixed and paraffin-embedded, was analyzed using a custom nCounter assay that detects 174 common sarcoma gene fusions. The results were compared to fluorescence in situ hybridization (FISH)/next-generation sequencing (NGS) and clinicopathological data.

Results

The nCounter assay was conclusive in 98.2% of cases, identifying 25 gene fusions with 82.5% accuracy, 76.6% sensitivity, and 100% specificity compared to FISH/NGS.

Conclusion

Although it does not identify all sarcoma fusions, especially for rare subtypes, the present nCounter assay is a rapid, affordable, and accurate tool for sarcoma diagnosis in resource-limited settings.

Remote web-based self-assessment of visual acuity versus ETDRS in patients with macular diseases: a method comparison study

BACKGROUND

Macular diseases (MD) lead to frequent clinic visits, involve time-consuming visual acuity (VA) measurements by professionals. Independent home measurements could improve efficiency. This study evaluates the agreement of a web-based test in MD compared to in-hospital measurements.

METHODS

Adults with MD were included at the University Medical Center Utrecht in March-July 2023. Users need a phone, computer, and 3m distance. The test uses Tumbling-E and triangles as optotypes. Primary outcome is the web-based vs. ETDRS Distance Visual Acuity (DVA). Secondary outcomes were test-retest variability (TRV), near visual acuity (NVA), and the Amsler grid. Outcomes were reported in mean differences and 95% Limits of Agreement (LoA).

RESULTS

89 eyes were included. The DVA mean difference was 0.03LogMAR(1.5 letters), SD0.17, LoA - 0.31;0.36LogMAR(-15.5;18 letters), TRV had a mean difference of 0.03(1.5 letters) SD0.14. The NVA mean difference was 0.13(6.5 letter) SD0.24, positive- and negative predictive values 0.93(95%CI = 0.82;0.98) and 0.71(95%CI = 0.51;0.86) respectively.

CONCLUSIONS

The agreement of the DVA web-based test is on par with Snellen line assessment and subpar to ETDRS. We showed that elderly can perform this test independently at home, providing a time- and cost-saving opportunity. Developments should focus on the NVA since it can be a valuable adjunct to MD follow-up.

TRIAL REGISTRATION

the Dutch Medical Ethical committee (Medisch Ethische Toetsingscommissie; METC NedMec) registration number: 22-879/DB. Approved at 27-09-2022.

Diagnostic accuracy of an antigen-based point-of-care test versus nucleic acid amplification testing for genital trichomoniasis among pregnant women attending antenatal care facilities in Zambia

BACKGROUND

Infection with Trichomonas vaginalis (TV) is the most prevalent curable sexually transmitted infection (STI) globally and is associated with prelabour rupture of membranes, preterm delivery, and low birthweight. Point-of-care (POC) testing for TV during pregnancy may facilitate rapid antenatal case detection and treatment. This study, part of the World Health Organization's global ProSPeRo study, aimed to evaluate the performance of OSOM® Trichomonas Rapid Test, an antigen-based POC test, against a reference nucleic acid amplification test (NAAT) among pregnant women in Zambia. We also assessed the operational characteristics and patient acceptability of the POC test, within the context of WHO's target product profiles for STI POC tests.

METHODS

We enrolled pregnant women attending four health centres in Nchelenge, Zambia, for antenatal care between 15 February and 26 May 2023. Vaginal swabs for the TV POC test and a reference NAAT (Aptima® Trichomonas vaginalis assay) were obtained. POC test results were read independently by two study staff members. Study staff filled a questionnaire on the operational characteristics of the POC test, and participants were asked about their willingness to wait for results.

RESULTS

Paired POC and reference test samples were collected from 1,015 participants. Overall, 23.0% (233/1015) tested positive for TV by NAAT, and 15.3% (155/1015) tested positive by the POC test, with three inconclusive results. The overall sensitivity and specificity of the POC test were 66.4% (95% confidence intervals [CI] 57.7-74.1%) and 99.6% (95% CI: 98.8-99.9%), respectively. Sensitivity was higher among those with TV-associated symptoms compared to those without (83.6% versus 60.4%, relative ratio 1.39, 95% CI 1.14-1.68). Inter-rater agreement was 99.7% (Cohen's Kappa 0.989). The study staff (n = 14) found the test easy to use and interpret, with most staff (12/14) reporting results were available within 25 min.

CONCLUSION

Overall, the TV POC test showed lower sensitivity than WHO's 85% target, but exceeded the 99% specificity target. Among symptomatic pregnant women, sensitivity nearly reached the WHO target. The assay was user-friendly, required minimal training, and delivered results quickly. Further studies are needed to determine the optimal antenatal settings for this technology.

TRIAL REGISTRATION

PACTR202302766902029.

Field evaluation of nanopore targeted next-generation sequencing to predict drug-resistant tuberculosis from native sputum in South Africa and Zambia

Rapid and comprehensive drug susceptibility testing (DST) is essential for diagnosing and treating drug-resistant tuberculosis effectively, and next-generation sequencing can be an effective genotypic DST method. We implemented and evaluated the performance of a nanopore targeted sequencing assay, called the Tuberculosis Drug Resistance Test (TBDR, Oxford Nanopore Diagnostics, Ltd., United Kingdom), which predicts drug resistance to 16 TB drugs, at a South African reference laboratory and a district diagnostic laboratory in Zambia. We compared the sequencing success rates between unprocessed and decontaminated sputum samples and determined the diagnostic accuracy against local DST (Xpert MTB/RIF Ultra, Xpert MTB/XDR, and BD BACTEC MGIT phenotypic DST). We prospectively sequenced 236 samples and have 148 samples with sequencing results from unprocessed and decontaminated sputum. We obtained successful sequencing results from 66.4% (94/148) unprocessed sputum samples and 75% (111/148) decontaminated samples. Sequencing success rates at the two sites differed, with 50.7% (36/71) successful sequencing results from unprocessed sputum in Zambia and 75.3% (58/77) in South Africa. Samples with "low" bacterial load, measured by Xpert MTB/RIF Ultra, tended to produce fewer successful sequencing results. TBDR sequencing predicted resistances in 48 samples, detecting resistance for rifampicin (n = 41) and isoniazid (n = 20), as well as 10 second-line drugs (n = 15). Sensitivity was variable compared to phenotypic DST, ranging from 33 (ethionamide) to 94% (rifampicin), while specificity remained above 90% for all drugs, except clofazimine. The TBDR assay can provide rapid, comprehensive genotypic DST. Technical and operational challenges need to be addressed for its broader implementation in high tuberculosis-burden settings.IMPORTANCEThis study illustrates the use of the Tuberculosis Drug Resistance Test (TBDR, Oxford Nanopore Diagnostics, Ltd., United Kingdom) as a rapid drug susceptibility testing (DST) approach for diagnosing drug-resistant TB in the high TB-burden countries of South Africa and Zambia. The TBDR assay predicts resistance to 16 TB drugs, including first- and second-line treatments. By implementing the TBDR assay in a national reference laboratory in South Africa and a district diagnostic laboratory in Zambia, we demonstrate how this technology can provide faster diagnostic results (days) compared to traditional phenotypic DST methods (~2 months), with adequate sensitivity. Missed resistances compared to phenotypic DST indicate that technical improvements are needed. Successful sequencing from unprocessed and decontaminated sputum samples at different sites suggests feasibility in diverse settings, though operational challenges remain. Implementing this rapid, comprehensive DST approach could enhance drug-resistant tuberculosis diagnosis and treatment, ultimately improving patient outcomes and helping to combat tuberculosis in high-burden regions.

Age-specific assessment of initial hemoglobin levels and shock index for predicting life-saving interventions in pediatric blunt liver and spleen injuries

This study aimed to evaluate the effectiveness of combining initial hemoglobin levels with the shock index for predicting the need for life-saving interventions (LSI) in pediatric patients with blunt liver and spleen injuries (BLSI), specifically tailored to different age groups. This was a multicenter retrospective cohort study of pediatric patients with BLSI in Japan. The area under the receiver operating characteristic curve (AUROC) were used to assess predictive accuracy. The study included 1,370 patients. LSI was required in 59 of 247 (23.9%) aged 1 to 6 years, 100 of 402 (24.9%) aged 7 to 12 years, and 125 of 297 (42.1%) patients aged 13 to 16 years. Within each specific age group, the predictability was categorized as fair and appeared higher than that of the entire cohort or when using either parameter alone. Notably, in the 1 to 6-year age group, the combined values showed the highest predictability, which was statistically superior to the shock index alone (AUROC of 0.770 vs. 0.671, P = 0.025). Tailoring initial hemoglobin levels and shock index to specific age groups enhances predictability of LSI in pediatric BLSI, showing a fair level of predictive accuracy.

Validation of the capnodynamic method to calculate mixed venous oxygen saturation in postoperative cardiac patients

BACKGROUND

Cardiac output and mixed venous oxygen saturation are key variables in monitoring adequate oxygen delivery and have typically been measured using pulmonary artery catheterisation. The capnodynamic method measures effective pulmonary blood flow utilising carbon dioxide kinetics in ventilated patients. Combined with breath-by-breath measurements of carbon dioxide elimination, a non-invasive approximation of mixed venous oxygen saturation can be calculated.

METHODS

This study primarily investigated the agreement between mixed venous oxygen saturation calculated using the capnodynamic method and blood gas analysis of mixed venous blood sampled via a pulmonary artery catheter in 47 haemodynamically stable postoperative cardiac patients. Both measurements were synchronised and performed during alveolar recruitment by stepwise changes to the level of positive end-expiratory pressure. Simultaneously, we studied the agreement between effective pulmonary blood flow and thermodilution cardiac output. The Bland-Altman method for repeated measurements and calculation of percentage error were used to examine agreement. Measurements before and after alveolar recruitment were analysed by a paired t test. The study hypothesis for agreement was a limit of difference of ten percentage points between mixed venous oxygen saturation using the capnodynamic algorithm vs. catheter blood gas analysis.

RESULTS

Capnodynamic calculation of mixed venous saturation compared to blood gas analysis showed a bias of -0.02 [95% CI - 0.96-0.91] % and limits of agreement at 8.8 [95% CI 7.7-10] % and - 8.9 [95% CI -10-- 7.8] %. The percentage error was < 20%. The effective pulmonary blood flow compared to thermodilution showed a bias of - 0.41 [95% CI - 0.55-- 0.28] l.min-1 and limits of agreement at 0.56 [95% CI 0.41-0.75] l.min-1 and - 1.38 [95% CI - 1.57--1.24] l.min-1. The percentage error was < 30%. Only effective pulmonary blood flow increased by 0.38 [95% CI 0.20-0.56] l.min-1 (p < 0.01) after alveolar recruitment.

CONCLUSIONS

In this study, minimal bias and limits of agreement < 10% between mixed venous oxygen saturation calculated by the capnodynamic method and pulmonary arterial blood gas analysis confirmed the agreement hypothesis in stable postoperative patients. The effective pulmonary blood flow agreed with thermodilution cardiac output, while influenced by pulmonary shunt flow.

An artificial intelligence tool that may assist with interpretation of rapid plasma reagin test for syphilis: Development and on-site evaluation

OBJECTIVES

The rapid plasma reagin (RPR) test, a traditional method for diagnosing syphilis and evaluating treatment efficacy, relies on subjective interpretation and requires high technical proficiency. This study aimed to develop and validate a user-friendly RPR-artificial intelligence (AI) interpretative tool.

METHODS

A dataset comprising 600 images of photographed RPR cards from 276 negative and 223 positive RPR samples was used for model development. The reference result was based on consistent interpretations by at least two out of three experienced laboratory personnel. Then an interpretative model was developed using deep learning algorithms and loaded into smartphones for on-site interpretation at two clinical centers from October 2023 to April 2024.

RESULTS

The model demonstrated an accuracy of 82·67% (95% CI 71·82%-90·09%) for reactive circles and 84·44% (95% CI 69·94%-93·01%) for non-reactive circles. In the field study, 669 specimens showed a sensitivity of 94·85% (95% CI 89·29%-97·73%), specificity of 91·56% (95% CI 88·78%-93·71%), and concordance of 92·23% (95% CI 89·87%-94·09%). The positive predictive value was 74·14% (95% CI 66·86%-80·33%) and negative predictive value was 98.59% (95% CI 96·98%-99·38%).

CONCLUSIONS

The tool assists in RPR interpretation standardization, enabling data traceability, and quality control for remote and underdeveloped areas.

External Validation of the Phoenix Sepsis Score in Children With Suspected Community-Acquired Sepsis

Importance

The novel Phoenix Sepsis Score and sepsis criteria were derived and validated using a multicountry dataset and proposed as a new definition for sepsis in children.

Objective

To externally validate the Phoenix Sepsis Score and sepsis criteria in a cohort of children hospitalized with suspected community-acquired sepsis.

Design, Setting, and Participants

This diagnostic study used data from the multicenter, multicountry Sepsis Epidemiology in Australian and New Zealand Emergency Departments (SENTINEL) study, collected from 2021 to 2023 and including 90-day follow-up. Children admitted to the hospital through 11 emergency departments in Australia and New Zealand and treated with parenteral antibiotics with either (1) a provisional diagnosis of sepsis or (2) treatment for sepsis (intravenous fluid bolus to treat poor perfusion) were included.

Exposure

Development of organ dysfunction over the first 24 hours of hospitalization.

Main Outcomes and Measures

The main outcomes were (1) in-hospital mortality and (2) death or requirement for extracorporeal life support (ECLS) within 72 hours of hospitalization.

Results

A total of 6232 children were included in the analysis, with a median (IQR) age of 2.1 (0.3-7.1) years, 3386 (54.1%) male, in-hospital mortality of 60 (1.0%), and death or ECLS within 72 hours in 36 (0.6%). In this population, the worst Phoenix Sepsis Score calculated over the first 24 hours of hospitalization had an area under the precision recall curve of 0.17 (95% CI, 0.07-0.28) for predicting in-hospital mortality and 0.23 (95% CI, 0.11-0.36) for predicting death or ECLS within 72 hours. Overall, 306 children (4.9%) met the Phoenix sepsis criteria, of whom 33 (10.8%) died in the hospital (nearly half of the total number who died) and 28 (9.2%) died or required ECLS within 72 hours. The Phoenix sepsis criteria had a sensitivity of 55.0% (95% CI, 41.6%-67.9%) and positive predictive value (PPV) of 10.8% (95% CI, 7.6%-14.9%) for in-hospital mortality and sensitivity of 77.8% (95% CI, 60.8%-89.9%) and PPV of 9.2% (95% CI, 6.2%-13.0%) for death or ECLS within 72 hours. Coagulation data for the calculation of the Phoenix Sepsis Score were missing in more than 85% of children.

Conclusions and Relevance

In this multicenter diagnostic study of children hospitalized with suspected sepsis, the Phoenix Sepsis Score and sepsis criteria had similar performance to the original derivation and validation cohorts. The small proportion of children meeting Phoenix sepsis criteria, missingness of data, timing of application, and lack of sensitivity for in-hospital mortality limit the clinical utility of the criteria.

Diagnostic accuracy of artificial intelligence for dental and occlusal parameters using standardized clinical photographs

INTRODUCTION

SmileMate (SmileMate, Dental Monitoring SAS, Paris, France) is an artificial intelligence (AI)-based Web site that uses intraoral photographs to assess patients' dental and orthodontic parameters and provide a report. This study aimed to investigate the ability of an AI assessment tool (SmileMate) for orthodontic and dental parameters.

METHODS

A United Kingdom-based prospective clinical study enrolled 35 participants in the study. The participants' occlusal and dental parameters were assessed, and standardized orthodontic photographs were taken and uploaded to the SmileMate Web site to produce an AI-generated assessment. A total of 19 parameters were evaluated: 9 orthodontic parameters and 10 dental parameters covering both soft and hard tissues. A crosstabulation for AI and clinician assessments was reported using Fisher exact tests. Cohen's kappa was calculated to provide an agreement between the gold standard (clinician assessment) and SmileMate (AI assessment). Finally, the sensitivity, specificity, and area under the curve were calculated.

RESULTS

Statistically significant differences between a direct in-person assessment and the SmileMate AI assessment were noted across 9 of the 19 parameters (P <0.05, Fisher exact test). The overall kappa value was fair (0.29), with a variety of agreements between AI and clinician assessments; the level of agreement ranged from poor in 2 parameters (lateral open bite and teeth fracture) to almost perfect for missing and retained teeth. The level of agreement ranged from slight to moderate for the other variables in this study. The overall sensitivity of the AI-generated assessments was 72%, and the specificity was 54%. The specificity of AI was very low for gingivitis and oral hygiene, indicating a very high probability of false-positive findings for those parameters.

CONCLUSIONS

The overall agreement between SmileMate and the clinician's assessment was slight to moderate. AI-generated assessments are inadequate for evaluating malocclusion.

Performance of Hearing Test Software Applications to Detect Hearing Loss

Importance

Hearing loss is common and may impact health and quality of life if not properly managed. It is diagnosed following formal audiological assessment, which may not be available or practical. Hearing test software applications (apps) may help identify people who might benefit from audiological assessment, but their diagnostic accuracy has been incompletely studied.

Objective

To measure and compare the validity and reliability of 2 commonly recommended apps (hearWHO and SHOEBOX) to detect moderately severe or greater hearing loss. Secondary objectives were to evaluate the apps' ability to detect less severe hearing loss and the diagnostic performance of 2 questionnaires for detecting both severities of hearing loss.

Design, Setting, and Participants

This prospective diagnostic accuracy study compared the hearWHO and SHOEBOX apps with a 4-frequency pure-tone average audiological assessment reference standard. All consenting English-speaking patients aged 18 years or older and referred for routine audiological assessment at a publicly funded health center in Calgary, Canada, were included between May 17, 2023, and March 12, 2024.

Main Outcome and Measures

The main outcome was the validity and reliability of 4 index tests, including the hearWHO app, SHOEBOX app, Revised Hearing Handicap Inventory-Screening (RHHI-S) questionnaire, and the Single-Item Self-Assessment (SISA) questionnaire, to detect moderate to severe hearing loss. All index test results were compared with an audiological assessment reference standard (hearing loss defined by a better ear hearing threshold of ≥50 dB [more severe denoted as HL50] or ≥20 dB [less severe denoted as HL20]). Test-retest reliability of the 2 apps and C statistics, sensitivity, specificity, and positive and negative predicted values of all index tests were measured.

Results

A total of 130 participants were recruited (median [IQR] age, 58 [47-67] years; 82 female [63.1%]). Complete data for each comparison ranged from 123 to 129 participants. The prevalence of HL50 was 16.3% (21 or 130 participants). Neither the hearWHO nor the SHOEBOX app had high test-retest reliability (all κ-values <0.80), with the SHOEBOX having a κ of 0.64 (95% CI, 0.48-0.79) and hearWHO having a κ of 0.32 (95% CI, 0.18-0.46). All C statistics for HL50 were less than 0.80. When testing for HL50, diagnostic performance for both apps was better for the second measurement than the first measurement or the mean. Sensitivity and specificity for the second measurement of SHOEBOX were 0.26 (95% CI, 0.09-0.51) and 1.00 (95% CI, 0.97-1.00), respectively, and for the second measurement of hearWHO, 0.67 (95% CI, 0.43-0.85) and 0.71 (95% CI, 0.62-0.79), respectively. Sensitivity and specificity for the RHHI-S were 0.76 (95% CI, 0.53-0.92) and 0.42 (95% CI, 0.32-0.52), respectively, and for SISA, 0.10 (95% CI, 0.01-0.30) and 0.90 (95% CI, 0.83-0.95), respectively. Using a less stringent diagnostic threshold with SHOEBOX increased sensitivity for HL50 to at least 95% while retaining a specificity of 47% to 54%. Sensitivity and specificity for both apps were higher for HL20.

Conclusions and Relevance

These findings suggest that both hearWHO and SHOEBOX have limited test-retest reliability, perhaps because of a learning effect. Both apps may be suitable if a sensitive strategy is desired for identifying people who may benefit from diagnostic audiological assessment, whereas the SHOEBOX app may be preferable if a specific strategy is desired. If neither app is available, the RHHI-S or the SISA could be used depending on whether sensitivity or specificity is desired.

Hematocrit and drug interference study of the new Arkray Glucocard S Onyx blood glucose meter meeting ISO 15197:2013 accuracy criteria - a contribution to patient safety

INTRODUCTION

Abnormal HCT levels and medications can cause inaccurate readings with blood glucose meters (BGM) in routine practice. The GLUCOCARD S onyx (ARKRAY Inc.) BGM was evaluated over a wide HCT range (20-70%) and its interaction with 27 common drug interferences, as per ISO15197:2013/EN ISO15197:2015, compared to clinical laboratory reference methods.

METHODS

Pooled venous blood samples from >200 participants were prepared with four HCT concentrations (20-70%) and three glucose concentrations (60-360 mg/dL). These were spiked with 27 potential interfering substances and tested using GLUCOCARD S onyx, Atellica CH (D-Glucose), and Sysmex XN-10 (HCT).

RESULTS

Stable performance meeting ISO15197:2013 criteria were observed at HCT concentrations of 20-25%, 42±2%, 55±2%, and 66-69% (Levene, one-way ANOVA, p>0.05). No significant differences in averages were found for most substances, except for Ascorbic acid (>6 mg/dL), Sodium (>180 mmol/L), and Xylose (>600 mg/dL). Ascorbic acid at 3 mg/dL and Sodium at 90 mmol/L showed no interference for normal to low glucose readings.  .

CONCLUSION

The GLUCOCARD S onyx BGM provided accurate glucose readings across a wide HCT range (20-70%) and under various medication conditions, enabling reliable diabetes management by patients and effective treatment by healthcare providers in complex clinical scenarios.

[68Ga]Ga-PSMA-617 PET/MRI for imaging patients suspected of hepatocellular carcinoma

PURPOSE

Radiolabeled probes targeting prostate-specific membrane antigen (PSMA) have been used in prostate cancer. Moreover, PSMA is also overexpressed on neovessels in hepatocellular carcinoma (HCC). This study aimed to preliminarily evaluate the diagnostic effectiveness of [68Ga]Ga-PSMA-617 PET/MRI for HCC.

METHODS

Patients suspected of HCC were prospectively enrolled in this single-center study (NCT05006326, 2021-08-16) to perform [68Ga]Ga-PSMA-617 PET/MRI, along with contrast enhanced CT (ceCT) or ceMRI. The main suspicious intrahepatic lesions were resected and pathologically verified. Visual evaluation of [68Ga]Ga-PSMA-617 PET/MRI images was performed on all lesions. Maximum standard uptake value (SUVmax), mean standard uptake value (SUVmean), tumor-to-liver ratio (TLR), tumor-to-blood ratio (TBR), and tumor-to-parotid ratio (TPR) were measured or calculated. The diagnostic efficiency of different modalities was summarized. PSMA expression was evaluated by immunohistochemistry and the correlation of PSMA expression and [68Ga]Ga-PSMA-617 uptake in HCC primary tumors was quantitatively analyzed.

RESULTS

A total of 12 patients (ten men and two women; mean age 58.75 ± 12.08 years) were included. Ten patients were diagnosed with HCC, 2 with intrahepatic cholangiocarcinoma (ICC), and 4 with hemangioma. The SUVmax, TLR, TBR, and TPR of HCC primary tumors were higher than those of ICC and hemangioma. The diagnostic accuracy of [68Ga]Ga-PSMA-617 PET/MRI for primary HCC was 82.4%. When combined with ceCT or ceMRI, the accuracy increased to 88.2%. A moderate correlation was observed between SUVmax and mean PSMA expression in HCC primary tumors (R = 0.788).

CONCLUSION

Utilizing a hybrid PET/MRI system to combine [68Ga]Ga-PSMA-617 PET/MRI with ceMRI is a promising one-stop solution for the accurate diagnosis of HCC.

TRIAL REGISTRATION

NCT05006326. Registered August 16, 2021, https://clinicaltrials.gov/study/NCT05006326 .

Value of clinical review for AI-guided deep vein thrombosis diagnosis with ultrasound imaging by non-expert operators

Deep vein thrombosis (DVT) carries high morbidity, mortality, and costs globally. Point of care ultrasound (POCUS) image acquisition by non-ultrasound-trained providers, supported by an AI-based guidance and remote image review system, is believed to improve the timeliness and cost-effectiveness of diagnosis. We examine a database of 381 patients with suspected DVT who underwent an AI-guided ultrasound scan by a non-ultrasound-trained nurse and an expert sonographer-performed standard compression ultrasound scan. Each AI-guided scan was reviewed remotely by blinded radiologists or blinded independent POCUS-certified American Emergency Medicine (EM) physicians. Remote reviewer and standard scan diagnoses were compared. The primary endpoint is AI-guidance system sensitivity with clinician review; secondary endpoints include specificity, positive predictive value, negative predictive value, image quality, inter-observer image quality, and vein compressibility agreement. Data was analysed through the bootstrapping method, bootstrapping with a second reader for each scan, and a majority voting system. Eighty percent (n = 304) of scans were of sufficient diagnostic quality. Radiologist reviewer sensitivity ranged from 90%-95%, specificity from 74%-84%, NPV from 98%-99%, PPV from 30%-42%, and potential expert-led ultrasound scans avoided from 39%-50%. Inter-observer agreement for image quality was 0.15 and for compressibility 0.61. EM reviewer sensitivity ranged from 95%-98%, specificity from 97%-100%, NPV was 99%, PPV from 81%-100%, and potential expert-led ultrasound scans avoided from 29%-38%. Inter-observer agreement for image quality was 0.59 and for compressibility 0.67. Diagnosing lower extremity DVT through AI-guided image acquisition with clinician review is feasible. Performance is influenced by reviewer expertise. We find potential positive impacts on health economics, including safely avoiding expert-led ultrasound scans.

Accuracy and costs of bedside methods for confirming nasoenteral feeding tube position: a diagnostic accuracy study

AIM

To analyze the accuracy and costs of bedside methods for confirming the position of a nasoenteral feeding tube newly inserted blindly by nurses.

DESIGN

Diagnostic accuracy study of three clinical methods (ultrasound, epigastric auscultation, and pH measurement) compared to radiography. The direct costs of each method used to confirm the positioning of the nasoenteral tube were also measured.

METHODS

Seventy-six adult patients underwent a total of 87 nasoenteral tube insertion procedures in hospital units located within the Northeast region of the State of São Paulo, Brazil. The clinical methods were conducted on all study participants in the specified sequence: ultrasound (as index test), followed by epigastric auscultation and pH measurement (also index tests).

RESULTS

The outcomes regarding the confirmation of the accurate positioning of the nasoenteral tube are as follows: ultrasonography demonstrated sensitivity and specific of 79.0% and 66.7%, respectively. Epigastric auscultation exhibited a sensitivity of 81.3% and specificity of 83.3%. The pH measurement method displayed sensitivity and specificity of 89.3% and 100% respectively. Additionally, in terms of estimated direct costs, the pH measurement method incurred a higher cost (USD $8.31) compared to the other methods, with a difference of USD $6.68.

CONCLUSIONS

Based on these results, X-ray examination remains the primary method for confirming the placement of nasoenteral tubes recently inserted blindly at the bedside. However, when considering the costs of the evaluated methods, it is advisable to consider the variations in expenses between non-radiological methods and X-ray examinations.

Accuracy analysis of cholesterol analyzer in detecting dyslipidemia in truck drivers

Cardiovascular diseases (CVD) are a leading cause of mortality and morbidity worldwide. Rapid diagnostic tools are crucial for timely intervention, especially in high-risk groups such as truck drivers. In Brazil, the Mission® test uniquely offers test strips for simultaneous measurement of total cholesterol (TC), high-density lipoprotein (HDL), triglycerides (TG), and low-density lipoprotein (LDL). This study evaluates the accuracy of the Mission® analyzer compared to laboratory testing for HDL-C, TG, and TC in truck drivers. A blinded cross-sectional study was conducted among truck drivers aged 30-64 in Campo Grande, Mato Grosso do Sul, Brazil. Spearman correlation, linear regression, and the Bland-Altman analyses were employed to compare lipid profile results between the Mission® analyzer and laboratory methods. A total of 108 samples were analyzed. For HDL, the Mission® analyzer showed a sensitivity of 0.88, a specificity of 0.67, and an area under the curve (AUC) of 0.77 (95% CI: 0.68-0.86). For TG, sensitivity and specificity were 0.96 and 0.98, respectively, with an AUC of 0.97 (95% CI: 0.93-1.0). For TC, the AUC was 0.87 (95% CI: 0.79-0.95). Bland-Altman analysis revealed biases of -4.5 for HDL, 12.4 for TC, and -42.8 for TG between Mission® and laboratory results. The Mission® analyzer demonstrates good accuracy for rapid dyslipidemia diagnosis and Framingham Global Risk Score calculation. It is a valuable tool for initial screening and risk assessment, confirmation with laboratory testing is recommended for definitive diagnosis and treatment planning.

Artificial Intelligence-Guided Lung Ultrasound by Nonexperts

Importance

Lung ultrasound (LUS) aids in the diagnosis of patients with dyspnea, including those with cardiogenic pulmonary edema, but requires technical proficiency for image acquisition. Previous research has demonstrated the effectiveness of artificial intelligence (AI) in guiding novice users to acquire high-quality cardiac ultrasound images, suggesting its potential for broader use in LUS.

Objective

To evaluate the ability of AI to guide acquisition of diagnostic-quality LUS images by trained health care professionals (THCPs).

Design, Setting, and Participants

In this multicenter diagnostic validation study conducted between July 2023 and December 2023, participants aged 21 years or older with shortness of breath recruited from 4 clinical sites underwent 2 ultrasound examinations: 1 examination by a THCP operator using Lung Guidance AI and the other by a trained LUS expert without AI. The THCPs (including medical assistants, respiratory therapists, and nurses) underwent standardized AI training for LUS acquisition before participation.

Interventions

Lung Guidance AI software uses deep learning algorithms guiding LUS image acquisition and B-line annotation. Using an 8-zone LUS protocol, the AI software automatically captures images of diagnostic quality.

Main Outcomes and Measures

The primary end point was the proportion of THCP-acquired examinations of diagnostic quality according to a panel of 5 masked expert LUS readers, who provided remote review and ground truth validation.

Results

The intention-to-treat analysis included 176 participants (81 female participants [46.0%]; mean [SD] age, 63 [14] years; mean [SD] body mass index, 31 [8]). Overall, 98.3% (95% CI, 95.1%-99.4%) of THCP-acquired studies were of diagnostic quality, with no statistically significant difference in quality compared to LUS expert-acquired studies (difference, 1.7%; 95% CI, -1.6% to 5.0%).

Conclusions and Relevance

In this multicenter validation study, THCPs with AI assistance achieved LUS images meeting diagnostic standards compared with LUS experts without AI. This technology could extend access to LUS to underserved areas lacking expert personnel.

Trial Registration

ClinicalTrials.gov Identifier: NCT05992324.

Added Value of Shaking Chills for Predicting Bacteremia in Patients with Suspected Infection

BACKGROUND

Detailed grading of chills is more useful for diagnosing bacteremia than simply classifying the presence or absence of chills. However, its value added to other clinical information has not been evaluated.

OBJECTIVE

To evaluate the value of adding chills grading to other clinical information compared to simply noting the presence or absence of chills for predicting bacteremia in patients with suspected infection.

DESIGN

Prospective observational study.

PARTICIPANTS

Adult patients admitted to two acute-care hospitals with suspected infection from April 2018 to March 2019.

MAIN MEASURES

Two types of categorization for chills were applied: "presence" or "absence" (dichotomized chills); and "no chills", "mild/moderate chills", and "shaking chills" (trichotomized chills). Three multivariable logistic regression models incorporating each of dichotomized chills, trichotomized chills, and C-reactive protein (CRP) with other clinical information were developed and compared. To assess the potential consequences of using each model to identify patients with high risk of bacteremia (i.e., requiring prompt intervention), we applied a cut-off point of an estimated probability of 60%. The number of patients with bacteremia correctly identified by each model was compared.

KEY RESULTS

Among the 2,013 patients, 327 (16.2%) were diagnosed with bacteremia. The three models showed comparable discrimination and calibration performance. At the 60% cut-off, the dichotomized chills model correctly identified 11 patients (3.4% [95% confidence interval (CI) 1.9-3.4] of patients with bacteremia). The trichotomized chills model and CRP model correctly identified an additional 15 patients (4.6% [95% CI 2.8-7.4]) and 2 patients (0.6% [95% CI 0.1-2.3]) with bacteremia, respectively.

CONCLUSIONS

Differentiating shaking chills in comparison with dichotomized chills for predicting bacteremia allowed the correct identification of an additional 4.6% of patients with bacteremia. Detailed grading of chills can be assessed without additional time, cost, or burden on patients and can be recommended in the routine history taking.

Head-to-Head Comparison Between Phosphatidylethanol Versus Indirect Alcohol Biomarkers for Diagnosis of MetALD Versus MASLD: A Prospective Study

BACKGROUND

The current subclassification of steatotic liver disease (SLD) relies on validated questionnaires, such as Alcohol Use Disorders Identification Test (AUDIT) and Lifetime Drinking History (LDH), which, while useful, are impractical and lack precision for their use in routine clinical practice. Phosphatidylethanol (PEth) is a quantitative, objective alcohol biomarker with high sensitivity and specificity.

AIMS

To assess the diagnostic accuracy of PEth for differentiating metabolic dysfunction and alcohol-associated liver disease (MetALD) from metabolic dysfunction-associated steatotic liver disease (MASLD) in a large, population-based, prospective, multiethnic cohort of individuals with overweight or obesity.

METHODS

This is a cross-sectional analysis of a prospective study including 374 adults with overweight or obesity residing in Southern California who had SLD as defined by MRI-PDFF ≥ 5%. The clinical research visit included medical history, biochemical and PEth testing, standardised validated questionnaires (including AUDIT and LDH), physical examination, and advanced imaging using MRI-PDFF and MRE.

RESULTS

Among 374 adults with SLD, the prevalence of MASLD, MetALD, and ALD was 90.1%, 6.4%, and 3.5%, respectively. PEth had a robust diagnostic accuracy in the detection of MetALD (AUROC 0.81, 95%CI 0.73-0.89) and the Youden cut-off was 25 ng/mL. In head-to-head comparative efficacy analysis, PEth was both statistically and clinically superior to all previously used indirect alcohol biomarkers for diagnosing MetALD, including aspartate aminotransferase/alanine aminotransferase ratio, mean corpuscular volume, gamma glutamyltransferase, and ALD/NAFLD index (p < 0.05).

CONCLUSIONS

PEth outperforms previously used non-invasive tests in differentiating MetALD from MASLD and has the potential to change clinical practice by enhancing the subclassification of SLD.

C-Reactive Protein-based Screening of People with Tuberculosis Symptoms: A Diagnostic Accuracy Study

Rationale: C-reactive protein (CRP)-based tuberculosis (TB) screening is recommended for people with HIV. However, its performance among people without HIV and in diverse settings is unknown. Objectives: In a multicountry study, we aimed to determine whether CRP meets the minimum accuracy targets (sensitivity ⩾ 90%, specificity ⩾ 70%) for an effective TB screening test. Methods: Consecutive outpatient adults with cough ⩾2 weeks from five TB endemic countries in Africa and Asia had baseline blood collected for point-of-care CRP testing and HIV and diabetes screening. Sputum samples were collected for Xpert MTB/RIF Ultra (Xpert) testing and culture. CRP sensitivity and specificity (5 mg/L cut-point) was determined in reference to sputum test results and compared by country, sex, and HIV and diabetes status. Variables affecting CRP performance were identified using a multivariate receiver operating curve regression model. Measurements and Main Results: Among 2,904 participants, of whom 613 (21%) had microbiologically confirmed TB, CRP sensitivity was 84% (95% confidence interval [CI], 81-87%) and specificity was 61% (95% CI, 59-63%). CRP accuracy varied geographically, with higher sensitivity in African countries (⩾91%) than Asian countries (64-82%). Sensitivity was higher among men than women (86% vs. 78%; difference, +8%; 95% CI, 1-15%) and specificity was lower among people with HIV than people without HIV (64% vs. 45%; difference, +19%; 95% CI, 13-25%). Receiver operating curve regression identified country and measures of TB disease severity as predictors of CRP performance. Conclusions: Overall, CRP did not achieve the minimum accuracy targets, and its performance varied by setting and in some subgroups, likely reflecting population differences in mycobacterial load.

PCR array analysis reveals a novel expression profile of ferroptosis-related genes in idiopathic pulmonary fibrosis

BACKGROUND

Idiopathic pulmonary fibrosis (IPF) is a chronic, irreversible, and fatal disease characterized by progressive interstitial lung fibrosis. Given its insidious onset and poor outcome, there is an urgent need to elucidate the molecular mechanisms underlying IPF and identify effective therapeutic targets and diagnosis and prognosis biomarkers. Ferroptosis is an iron-dependent form of programmed cell death that occurs as lipid peroxides accumulate. Growing evidence suggests that ferroptosis is important in IPF.

METHODS

Human ferroptosis PCR array was performed on IPF and control lung tissue. The differentially expressed ferroptosis-related genes (DE-FRGs) were identified, underwent functional enrichment analyses, protein-protein interaction network construction, and potential drug target prediction. The DE-FRGs were validated and their value as diagnostic and prognostic blood biomarkers were evaluated using the Gene Expression Omnibus dataset GSE28042.

RESULTS

The array identified 13 DE-FRGs. Gene Ontology enrichment and Kyoto Encyclopedia of Genes and Genomes pathway analyses revealed that the DE-FRGs were mainly related to iron ion transport, blood microparticles, and oxidoreductase activity, and were involved in porphyrin metabolism, necroptosis, and the p53 signaling pathway in addition to ferroptosis. The 13 DE-FRGs were analyzed using the Drug-Gene Interaction Database to explore novel IPF therapeutic agents, yielding 42 potential drugs. Four DE-FRGs (BBC3, STEAP3, EPRS, SLC39A8) in the peripheral blood of IPF patients from the GSE28042 dataset demonstrated the same expression pattern as that observed in the lung tissue array. The receiver operating characteristic analysis demonstrated that the area under the curve of STEAP3 and EPRS were > 0.75. The survival analysis demonstrated that STEAP3 and EPRS were significantly different between the IPF and control groups.

CONCLUSIONS

The FRG expression profiles in IPF and control lung tissue were characterized. The findings provided valuable ideas to elucidate the role of ferroptosis in IPF and aided the identification of novel IPF therapeutic targets and biomarkers.

The hypotension prediction index versus mean arterial pressure in predicting intraoperative hypotension: A clinical perspective

BACKGROUND

The hypotension prediction index (HPI) predicts hypotension, with randomised trials showing a significant reduction in hypotension-related metrics. However, the reliability of previous validation studies is debated, and it's unclear if mean arterial pressure (MAP) can be used interchangeably with HPI.

OBJECTIVES

This study compared the effectiveness of HPI versus MAP thresholds in predicting intraoperative hypotension, focusing on three clinically relevant metrics: time from alert to event, positive predictive value (PPV), and proportion of missed hypotensive events.

DESIGN

Prospective observational study conducted between 2018 and 2020.

SETTING

Single-centre, academic hospital in the Netherlands.

PARTICIPANTS

Adults scheduled for elective non-cardiac surgery lasting over two hours. Of the 105 eligible patients, 91 had sufficient data for analysis.

MAIN OUTCOME MEASURES

The primary outcome was the time-to-hypotensive event intervals predicted by HPI popup alerts (≥85 for ≥40 s) and MAP-alerts (70-75 mmHg). Secondary analyses examined differences between these predictors regarding the PPV and missed event rates, as well as the difference in these metrics between instant HPI-85 alerts and the six MAP-alerts.

RESULTS

The largest time-to-event difference was seen between HPI-85 popup and MAP-70 alerts, with a gain of 0.58 (95% confidence interval (CI), 0.57 to 0.58) min, favouring HPI. Higher MAP thresholds reduced this time difference, but worsened PPV values, with 20.5 (95% CI, 20.3 to 20.6)% at MAP-75 compared to 55.6 (95% CI, 55.4 to 55.8)% for HPI-85 popups. Missed event proportions were similar: between one to three percent. Instant HPI-85 and MAP-72 alerts showed comparable performance, but both had suboptimal PPV values around 30%. However, adding a 40-s time-dependence to MAP's alert definition levelled the differences across the three evaluated metrics, aligning more closely with HPI-85 popup alerts.

CONCLUSIONS

Using HPI-85 popup alerts does not provide additional prediction time over MAP-alerts in the 70 to 75 mmHg range, but they may be preferred due to higher PPV values. Instant HPI-85 and MAP-alerts perform similarly, with MAP-72 being closest, though these alerts more frequently occur regardless of subsequent hypotension with the potential to introduce unnecessary treatment. Adding a 40-s time-dependence to MAP-alerts to match the HPI popup characteristic eliminates distinctions between prediction time and missed events, while maintaining the higher PPV. However, whether 40sec-MAP-alerts are clinically equivalent remains to be determined in prospective clinical trials.

TRIAL REGISTRATION

Clinicaltrials.gov (NCT03795831) on 10 January 2019. https://clinicaltrials.gov/study/NCT03795831.

Automatic Detection of Radiographic Alveolar Bone Loss in Bitewing and Periapical Intraoral Radiographs Using Deep Learning Technology: A Preliminary Evaluation

Background/Objective: Periodontal disease is a prevalent inflammatory condition affecting the supporting structures of teeth, with radiographic bone loss (RBL) being a critical diagnostic marker. The accurate and consistent evaluation of RBL is essential for the staging and grading of periodontitis, as outlined by the 2017 AAP/EFP Classification. Advanced tools such as deep learning (DL) technology, including Denti.AI, an FDA-cleared software utilizing convolutional neural networks (CNNs), offer the potential for enhancing diagnostic accuracy. This study evaluated the diagnostic accuracy of Denti.AI for detecting RBL in intraoral radiographs. Methods: A dataset of 39 intraoral radiographs (22 periapical and 17 bitewing), covering 316 tooth surfaces (123 periapical and 193 bitewing), was selected from a de-identified pool of 500 radiographs provided by Denti.AI. RBL was assessed using the 2017 AAP/EFP Classification. A consensus panel of three board-certified dental specialists served as the reference standard. Performance metrics, including sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), accuracy, and mean absolute error (MAE), were calculated. Results: For periapical radiographs, Denti.AI achieved a sensitivity of 76%, specificity of 86%, PPV of 83%, NPV of 80%, and accuracy of 81%, with an MAE of 0.046%. For bitewing radiographs, sensitivity was 65%, specificity was 90%, PPV was 88%, NPV was 70%, and accuracy was 76%, with an MAE of 0.499 mm. Conclusions: Denti.AI demonstrated clinically acceptable performance in detecting RBL and shows potential as an adjunctive diagnostic tool, supporting clinical decision-making. While performance was robust for periapical radiographs, further optimization may enhance its accuracy for bitewing radiographs.

Development and Evaluation of an AI-Assisted Answer Assessment (4A) for Cognitive Assessments in Nursing Education

Artificial intelligence (AI) can potentially enhance cognitive assessment practices in maternal and child health nursing education. Objectives: To evaluate the reliability, accuracy and precision, and external validity of an AI-assisted answer assessment (4A) program for cognitive assessments in nursing education. Methods: This study is a validation study. Initially, 170 nursing students from northern Thailand participated, with 52 randomly selected for detailed testing. Agreement testing between the 4A program and human experts was conducted using the intraclass correlation coefficient (ICC). Accuracy and precision testing compared 4A scores with human expert assessments via the McNemar test. External validation involved 138 participants to compare the 4A program's assessments against national examination outcomes using logistic regression. Results: Results indicated a high level of consistency between the 4A program and human experts (ICC = 0.886). With an accuracy of 0.808 and a precision of 0.913, compared to the human expert's accuracy of 0.923 and precision of 1.000. The McNemar test (χ2 = 0.4, p = 0.527) showed no significant difference in evaluation performance between AI and human experts. Higher scores on the 4A program significantly predicted success in the national nursing examination (OR: 1.124, p = 0.031). Conclusions: The 4A program demonstrates potential in reliably assessing nursing students' cognitive abilities and predicting exam success. This study advocates for the continued integration of AI in educational assessments and the importance of refining AI systems to better align with traditional assessment methods.

Diagnostic Accuracy of Golgi Protein 73 (GP73) for Liver Fibrosis Staging in Metabolic Dysfunction-Associated Steatotic Liver Disease: A Scoping Review and Cohort Study

Background/Objectives: Golgi protein 73 (GP73) is a transmembrane protein expressed by epithelial cells of the bile duct in the normal liver. High serum levels of GP73 have been detected in patients with acute or chronic liver diseases, MASLD, and its measurement has been suggested as a potential biomarker for liver fibrosis staging. We evaluated the utility of GP73 in the diagnosis of MASLD, MASH, and for liver fibrosis staging. Methods: We performed a literature scoping review to map the current evidence about the accuracy of GP73 in patients with MASLD. We searched in Medline and EMBASE for English studies reporting an AUC value of GP73 in diagnosing MASLD and MASH and evaluating GP73 for fibrosis staging. A narrative synthesis of the evidence was conducted. Moreover, we performed an observational study including 84 patients with MASLD, of which 60 were biopsy-confirmed MASH, and different liver fibrosis stages, and 15 healthy controls. Serum GP73 levels were determined using a chemiluminescent assay and reported as mean and standard deviation (SD). Sensitivity (SE), specificity (SP), the area under the receiver operating characteristic (AUROC) curve, and the optimal cut-off value were calculated. Data were considered statistically significant when p < 0.05. Results: Available studies evaluating GP73 in MASLD reported the ability to discriminate MASH from simple steatosis and distinguish patients at different fibrotic stages, but the evidence is still scarce. Our experimental study showed that the serum levels of GP73 were 30 ± 12 ng/mL in MASLD and 32 ± 12 ng/mL in MASH patients and were statistically higher than those of the control group (19 ± 30 ng/mL), increasing from liver fibrosis stage F0 to F4. GP73 levels were significantly higher in patients with significant and advanced fibrosis than controls and no significant fibrosis (p > 0.05). ROC analysis demonstrated that serum GP73 had a good diagnostic potential for MASLD (AUROC 0.85; SE 90%; SP 73%), MASH (AUROC 0.75; SE 82%; SP64%), and significant fibrosis (AUROC 0.7; SE 56%; SP 79%) and was better than other biomarkers for chronic liver diseases. Conclusions: Serum GP73 could support clinicians in the evaluation of patients with MASH and significant fibrosis.

Predicting extubation failure in preterm infants using lung ultrasound: a diagnostic accuracy study

OBJECTIVE

To determine the accuracy of pre-extubation lung ultrasound (LUS) to predict reintubation in preterm infants born <32 weeks' gestation.

DESIGN

Prospective diagnostic accuracy study.

SETTING

Two neonatal intensive care units.

METHODS

Anterior and lateral LUS was performed pre-extubation. The primary outcome was the accuracy of LUS scores (range 0-24) to predict reintubation within 72 hours. Secondary outcomes were accuracy in predicting (1) reintubation within 7 days, (2) reintubation stratified by postnatal age and (3) accuracy of lateral imaging only (range 0-12). Pre-specified subgroup analyses were performed in extremely preterm infants born <28 weeks' gestation. Cut-off scores, sensitivities and specificities were calculated using receiver operating characteristic analysis and reported as area under the curves (AUCs).

RESULTS

One hundred preterm infants with a mean (SD) gestational age of 27.4 (2.2) weeks and birth weight of 1059 (354) g were studied. Thirteen were subsequently reintubated. The AUC (95% CI) of the pre-extubation LUS score for predicting reintubation was 0.63 (0.45-0.80). Accuracy was greater in extremely preterm infants: AUC 0.70 (0.52-0.87) and excellent in infants who were <72 hours of age at the time of extubation: AUC 0.90 (0.77-1.00). Accuracy was poor in infants who were >7 days of age. Lateral imaging alone demonstrated similar accuracy to scanning anterior and lateral regions.

CONCLUSIONS

In contrast to previous studies, LUS was not a strong predictor of reintubation in preterm infants. Accuracy is increased in extremely preterm infants. Future research should focus on infants at highest risk of extubation failure and consider simpler imaging protocols.

TRIAL REGISTRATION NUMBER

Australian New Zealand Clinical Trials Registry: ACTRN12621001356853.

Effect of blood alcohol on the diagnostic accuracy of glial fibrillary acidic protein and ubiquitin carboxy-terminal hydrolase L1 for traumatic intracranial hemorrhage: A TRACK-TBI study

BACKGROUND

Acute intoxication is common in patients evaluated for traumatic brain injury (TBI). However, the effect of elevated blood alcohol levels (BALs) on the diagnostic accuracy of FDA-cleared biomarkers for evaluating traumatic intracranial injury on computed tomography (CT) scan, namely, glial fibrillary acidic protein (GFAP) and ubiquitin carboxy-terminal hydrolase L1 (UCH-L1), has not been well studied.

METHODS

We investigated the effect of significantly elevated (>300 mg/dL) and modestly elevated BAL (81-300 mg/dL) at emergency department presentation on the diagnostic accuracy of GFAP and UCH-L1 for predicting a positive CT in patients presenting to 18 U.S. Level I trauma centers within 24 h of TBI as part of the prospective, Transforming Research and Clinical Knowledge in TBI (TRACK-TBI) study. Plasma GFAP and UCH-L1 were measured using Abbott i-STAT Alinity and ARCHITECT assays. Discrimination was evaluated using the area under the receiver operating characteristic curve (AUC).

RESULTS

Of 2320 TRACK-TBI participants studied, 54 (2.3%), 332 (14.3%), 1209 (52.1%), and 725 (31.3%) had significantly elevated BAL, modestly elevated BAL, nonelevated BAL (0-80 mg/dL), and no BAL available, respectively; 48.3% of the cohort had a positive brain CT. Those with significantly elevated and modestly elevated BAL were more likely to have positive CT (61.1% and 60.5% vs. 46.9% and 44.0%) and had higher plasma GFAP and UCH-L1 levels than those with nonelevated BAL and no BAL available. The AUC of GFAP and UCH-L1 combined for predicting CT positivity was higher in those with significantly elevated BAL (0.949) than those with modestly elevated BAL (0.858), nonelevated BAL (0.849), and no BAL available (0.883).

CONCLUSIONS

Modestly and significantly elevated BAL does not lower the diagnostic accuracy of GFAP and UCH-L1 for predicting traumatic intracranial injury on CT. These biomarkers may be useful in decreasing avoidable brain CT imaging in persons with acute alcohol intoxication.

A Bayesian latent class estimation of the diagnostic accuracy of clinical examination and laboratory assays to identify bovine ephemeral fever virus infection in South African cattle

Bovine ephemeral fever (BEF) is an economically important vector-borne viral disease of cattle and water buffalo in Africa, Australia and parts of Asia. The control of BEF is centred around vaccination, and therefore accurate, early identification of disease outbreaks are key to minimize its economic and welfare impact. In Africa, control programs are hampered by limited diagnostic capabilities and poor infrastructure for rapid transportation of diagnostic specimens. The primary objective of this study was to estimate the sensitivity (Se) and specificity (Sp) of four tests, namely clinical examination by a veterinarian, virus isolation and two different conventional PCR assays, to identify an acute bovine ephemeral fever virus (BEFV) infection in diseased, naturally infected South African cattle, without the assumption of a reference standard. Samples and data were collected from cattle with clinical signs suggestive of BEF rather than a random sample of cattle. A case was categorised as clinical examination positive if the examining veterinarian considered acute BEFV-infection as the most likely aetiology. Virus isolation was performed using the buffy coat of heparin blood samples on baby hamster kidney cell cultures, evaluating cytopathic effect and confirming virus morphology by transmission electron microscopy. PCR was performed using two previously published protocols: The Ephemerovirus L-gene PCR (targeting the RNA-dependent RNA polymerase gene) and a BEFV G-gene PCR (targeting the neutralising G1 epitope of the glycoprotein). A single population, four test Bayesian latent class model with conditional dependence between the two PCR assays was implemented. The prevalence of BEFV-infection was high in this study population of clinical suspects at 67 %, (95 % Probability Interval (PI) 52 %; 81 %). Clinical examination provided a reasonable indication of acute BEFV infection (Se of 86 % (PI 77 %; 93 %) and Sp of 67 % (PI 52 %; 82 %)). Virus isolation was the most specific (99 % (PI 97 %; 100 %)), but least sensitive assay (30 % (PI 20 %; 44 %)). Of the two conventional PCRs, the L-gene PCR outperformed the G-gene PCR: The L-gene Se was 64 % (PI 51 %; 76 %) and Sp 96 % (PI 84 %; 100 %) compared to Se of 50 % (PI 38 %; 61 %) and Sp of 89 % (PI 75 %; 98 %) for the G-gene. While the laboratory assays presented excellent positive predictive values within this high disease prevalence population, the poor negative predictive values limit their usefulness to field veterinarians attempting to exclude BEF as diagnosis. Novel pen-side diagnostics should be developed due to the limitations of currently available assays and infrastructure constraints prevalent in Africa.

Validation of the alcohol use disorders identification test in a Danish hospital setting

BACKGROUND

Early identification of potential alcohol-problems is central for timely intervention and treatment referral. The Alcohol Use Disorders Identification Test (AUDIT) and AUDIT-Consumption (AUDIT-C) serve as globally recognized and validated screening tools for this purpose. We aimed to evaluate the diagnostic validity of internationally recommended AUDIT cut-off scores ≥ 8, ≥16, ≥ 20, and AUDIT-C cut-off scores ≥ 4, ≥5 using the Danish language versions of questionnaires in a hospital setting.

METHODS

Questionnaire data were collected from 2/15/2023, to 4/27/2023 at the Department of Gastroenterology and Hepatology, Odense University Hospital, Denmark. We tested the World Health Organization's recommended AUDIT cut-offs: ≥8 for hazardous use, ≥ 16 suggestive of dependence, ≥ 20 high likelihood of dependence, along with AUDIT-C ≥ 4 and ≥ 5 using the following reference standard: Danish low-risk drinking guidelines (≤ 10 standard drinks/week) for hazardous use and self-reported ICD-10 alcohol dependence criteria for alcohol dependence. Analyses included ROC curves, AUC, sensitivity, specificity, predictive values, and agreement.

RESULTS

Three hundred patients participated, mean age 52 years (SD 17.4, median 54) and 51.3% males. Mean AUDIT score was 4.5 (SD 5.8, median 3) with fourteen (4.7%) meeting at least three self-reported ICD-10 criteria for alcohol dependence. The prevalence of hazardous use was 10.7%. AUDIT ≥ 8 exhibited a sensitivity of 56% (95% CI 40.6-73.6) and specificity 91% (95% CI 87.8-94.5) for detecting hazardous use. Against at least three self-reported ICD-10 criteria for alcohol dependence, AUDIT cut-off ≥ 16 showed a sensitivity of 85% (95% CI 66.1-98.2) with 97% specificity (95% CI 96.0-99.2), while cut-off ≥ 20 had a sensitivity of 71% (95% CI 49.2-91.6) with 99% specificity (95% CI 98.1-99.9). The AUDIT-C cut-offs ≥ 4 and ≥ 5 exhibited low positive predictive values in detecting hazardous use (30.8% for ≥ 4 and 36.8% for ≥ 5) and dependence (13.5% for ≥ 4 and 18.4% for ≥ 5) and demonstrated a specificity ranging from 68.5 to 82.1% with negative predictive values from 98.2 to 100%.

CONCLUSION

In Danish gastroenterology and hepatology departments, the AUDIT and AUDIT-C may be used to identify patients who are unlikely to have an alcohol problem, while positive screen results should be carefully considered and followed by more exhaustive assessment.

An Assessment of Deep Learning's Impact on General Dentists' Ability to Detect Alveolar Bone Loss in 2D Intraoral Radiographs

Background/Objective: Deep learning (DL) technology has shown potential in enhancing diagnostic accuracy in dentomaxillofacial radiology, particularly for detecting carious lesions, apical lesions, and periodontal bone loss. However, its effect on general dentists' ability to detect radiographic bone loss (RBL) in clinical practice remains unclear. This study investigates the impact of the Denti.AI DL technology on general dentists' ability to identify bone loss in intraoral radiographs, addressing this gap in the literature. Methods: Ten dentists from the university's dental clinics independently assessed 26 intraoral radiographs (periapical and bitewing) for bone loss using a Likert scale probability index with and without DL assistance. The participants viewed images on identical monitors with controlled lighting. This study generated 3940 data points for analysis. The statistical analyses included receiver operating characteristic (ROC) curves, area under the curve (AUC), and ANOVA tests. Results: Most dentists showed minor improvement in detecting bone loss on periapical radiographs when using DL. For bitewing radiographs, only a few dentists showed minor improvement. Overall, the difference in diagnostic accuracy between evaluations with and without DL was minimal (0.008). The differences in AUC for periapical and bitewing radiographs were 0.031 and -0.009, respectively, and were not statistically significant. Conclusions: This study found no statistically significant improvement in experienced dentists' diagnostic accuracy for detecting bone loss in intraoral radiographs when using Denti.AI deep learning technology.

Provocative tests of anal sphincter function correlate with symptoms and subtypes of faecal incontinence

OBJECTIVES

High-resolution anorectal manometry (HRAM) is the established investigation in faecal incontinence (FI). However, provocative tests (functional lumen imaging probe (FLIP) and anal acoustic reflectometry (AAR)) have been proposed as alternatives. This study uniquely explores all three methods in correlation with FI symptoms and subtypes.

METHODS

This was a prospective observational study of patients with FI attending a tertiary pelvic floor unit between August 2022 and January 2024. Patients underwent HRAM, FLIP and AAR with the order randomised. FI severity was assessed with the Vaizey score and quality-of-life with the Manchester Health Questionnaire.

RESULTS

40 patients (39 women, median age: 62 (range: 38-85)) were recruited with 27 (67.5%) reporting urge FI, 8 (20%) mixed and 5 (12.5%) passive incontinence. FLIP squeeze measurements correlated with the Vaizey score, including incremental squeeze pressure at 40 mL (rs=-0.412; p=0.008) and 50 mL (rs=-0.414; p=0.009) and the pressure-diameter volume loop at 50 mL (rs=-0.402; p=0.011). Incremental squeeze opening pressure with AAR correlated with the Vaizey score (rs=-0.339; p=0.032). There was no correlation between symptom severity and HRAM parameters, or any parameter and quality-of-life scores. Resting parameters with all three modalities were lower in passive FI: mean resting pressure (HRAM; p=0.010), yield pressure (FLIP; p=0.031) and opening pressure (AAR; p=0.006). With FLIP, there was a trend towards reduced squeeze function in the urge group (pressure-diameter volume loop at 50 mL; p=0.295).

CONCLUSIONS

FLIP and AAR correlate better with FI symptoms compared with HRAM. Therefore, these provocative tests could be used to guide the management of FI in prospective studies.

Validity, Accuracy, and Safety Assessment of an Aerobic Interval Training Using an App-Based Prehabilitation Program (PROTEGO MAXIMA Trial) Before Major Surgery: Prospective, Interventional Pilot Study

BACKGROUND

Major surgery is associated with significant morbidity and a reduced quality of life, particularly among older adults and individuals with frailty and impaired functional capacity. Multimodal prehabilitation can enhance functional recovery after surgery and reduce postoperative complications. Digital prehabilitation has the potential to be a resource-sparing and patient-empowering tool that improves patients' preoperative status; however, little remains known regarding their safety and accuracy as medical devices.

OBJECTIVE

This study aims to test the accuracy and validity of a new software in comparison to the gold-standard electrocardiogram (ECG)-based heart rate measurement.

METHODS

The PROTEGO MAXIMA trial was a prospective interventional pilot trial assessing the validity, accuracy, and safety of an app-based exercise program. The Prehab App calculates a personalized, risk-stratified aerobic interval training plan based on individual risk factors and utilizes wearables to monitor heart rate. Healthy students and patients undergoing major surgery were enrolled. A structured risk assessment was conducted, followed by a 6-minute walking test and a 37-minute supervised interval session. During the exercise, patients wore app-linked wearables for heart rate and distance measurements, which were compared with standard ECG and treadmill measurements. Safety, accuracy, and usability assessments included testing alarm signals, while the occurrence of adverse events served as the primary and secondary outcome measures.

RESULTS

A total of 75 participants were included. The mean heart rate differences between wearables and standard ECG were ≤5 bpm (beats per minute) with a mean absolute percentage error of ≤5%. Regression analysis revealed a significant impact of the BMI (odds ratio 0.90, 95% CI 0.82-0.98, P=.02) and Timed Up and Go Test score (odds ratio 0.12, 95% CI 0.03-0.55, P=.006) on the accuracy of heart rate measurement; 29 (39%) patients experienced adverse events: pain (5/12, 42%), ECG electrode-related skin irritations (2/42, 17%), dizziness (2/42, 17%), shortness of breath (2/42, 17%), and fatigue (1/42, 8%). No cardiovascular or serious adverse events were reported, and no serious device deficiency was detected. There were no indications of clinically meaningful overexertion based on laboratory values measured before and after the 6-minute walking test and exercise. The differences in means and ranges were as follows: lactate (mmol/l), mean 0.04 (range -3 to 6; P=.47); creatinine kinase (U/l), mean 12 (range -7 to 43; P<.001); and sodium (mmol/l), mean -2 (range -11 to 12; P<.001).

CONCLUSIONS

The interventional trial demonstrated the high safety of the exercise program and the accuracy of heart rate measurements using commercial wearables in patients before major surgery, paving the way for potential remote implementation in the future.

TRIAL REGISTRATION

German Clinical Trials Register DRKS00026985; https://drks.de/search/en/trial/DRKS00026985 and European Database on Medical Devices (EUDAMED) CIV-21-07-0307311.

INTERNATIONAL REGISTERED REPORT IDENTIFIER (IRRID)

RR2-10.1136/bmjopen-2022-069394.

Histology of the Upper Gastrointestinal Tract, Morphometry and Lymphocyte Subpopulations of the Duodenal Mucosa: Insights from Healthy Individuals

The upper oesophagogastrointestinal (UEGI) tract histology, intestinal morphometry and lymphocyte subpopulations of healthy people is scarcely known. In research studies of inflammation involving the UEGI tract, there is a lack of adequate healthy controls. Aims: To evaluate the histology of the UEGI tract and the duodenal lymphocyte subpopulations of healthy volunteers and patients with gastroesophageal reflux disease (GERD), the latter to assess if it could replace healthy subjects. Healthy individuals were excluded if they had symptoms, comorbidities, pregnancy, toxics, medications or abnormal blood analysis. Subjects in both groups with abnormal duodenal intraepithelial lymphocyte (IEL) counts were also excluded. A total of 280 subjects were assessed, and 37 were included (23 healthy and 14 with GERD). The GERD group showed a higher IEL count (median [IQR]: 19.5 [17-22]), than healthy group: (15 [12-18]), p = 0.004. Eosinophils, mast cells and intestinal morphometry were similar in both groups. In the lamina propria, CD4+ T cells decreased (p = 0.008), and CD8+ T cells increased (p = 0.014). The total innate lymphoid cells (ILC) and CD3- cells decreased (p = 0.007) in GERD group compared to healthy controls. At the intraepithelial level, NKT cells increased (p = 0.036) and ILC3 decreased (p = 0.049) in the GERD group. This is the first study to comprehensively map the histology, morphometry and duodenal subpopulations of healthy volunteers to help define a "gold standard" of normality. The differences found between both groups suggest that, whenever possible, healthy subjects should be included in research studies. Alternatively, we can consider a well-defined homogenous group with GERD to serve as the control group.

Closing the gap while standing still: clinimetric properties of a low-cost balance platform and a user-friendly app for posturography

Background

The Wii Balance Board (WBB) is used as a rehabilitation tool for balance or strength interventions and posturography in balance tasks. Nonetheless, implementation of posturography using the WBB in a clinical setting is hampered by required technical skills for signal processing to obtain meaningful balance measures. Therefore, this study aims to evaluate the concurrent validity and test-retest reliability of a WBB to measure center of pressure (COP) parameters and to provide an easy-to-use web application to improve implementation of posturography in clinical practice.

Methods

A cross-sectional study was carried out including 30 healthy adults who performed repeated balance tasks including single and double leg standing still with eyes open or eyes closed. A WBB on top of a laboratory-grade force plate synchronously measured COP. Parameters based on COP displacement were calculated, including standard deviation of displacement, velocity, pathlength and 95% predicted ellipse area.

Results

The concurrent validity of the WBB to measure COP in quiet standing still tasks was excellent for all parameters (Intraclass Correlation Coefficient (ICC) > 0.900, p < 0.001), apart from medio-lateral velocity (ICC = 0.571, p = 0.090 to ICC = 0.711, p = 0.057). For the single leg balance tasks, across the two measurements, all WBB COP derived parameters showed excellent correlations with COP parameters derived from a laboratory-grade force plate (ICC > 0.95, p < 0.001). Test-retest reliability of the WBB was poor (ICC below 0.5) to occasionally good (ICC between 0.75 to 0.90) for the COP parameters from quiet standing balance tasks. Comparable reliability was found for the repeated measurements of single leg standing still. Power spectra analysis of both force plates revealed larger measurement error by the WBB in medio-lateral direction in tasks requiring minimal postural adjustments.

Conclusion

The WBB revealed excellent concurrent validity with a laboratory-grade force plate for balance tasks on a single leg or two legs for most COP parameters. The reliability was poor to moderate for most tasks, however comparable to the findings from the laboratory grade force plate. An open-source web application, employing R Shiny, was created to provide a tool to analyse COP parameters. Hereby, it was demonstrated that open-source scientific tools may help researchers to bridge the gap between scientific findings and clinical use of posturography.

Validity and Reliability of the Swedish Version of the Gugging Swallowing Screen for use in Acute Stroke Care

The purpose of this study was to cross-culturally validate the Swedish version of the Gugging Swallowing Screen (GUSS-S) for use in the acute phase of stroke. Further, to evaluate the inter-rater reliability between different healthcare professionals. GUSS was translated into Swedish using a forward-backward method followed by expert rating to obtain content validity. For criterion validity, the GUSS-S score was compared with Flexible Endoscopic Evaluation of Swallowing (FEES) assessed with the Penetration-Aspiration Scale (PAS) in acute stroke patients (≤ 96 h after stroke onset). Convergent validity was calculated by comparison with the Functional Oral Intake Scale (FOIS) as per the comprehensive FEES assessment, the Standardized Swallowing Assessment (SSA), and the National Institutes of Health Stroke Scale (NIHSS). To evaluate inter-rater reliability, a nurse and a speech-language pathologist (SLP) independently assessed 30 patients. In total, 80 patients (32 women, median age 77 years (range 29-93) were included, mean 1.7 ± 0.9 days after admission. With a cut-off value of 14 points, the GUSS-S identified aspiration with a sensitivity of 100% and a specificity of 73% (area under the curve: 0.87, 95% CI 0.78-0.95). Spearman rank correlation showed very strong correlation between the GUSS-S and PAS (rs=-0.718, P = < 0.001) and FOIS (rs=0.720, P = 0.001) and strong correlation between the GUSS-S and SSA (rs=0.545, P = < 0.001) and NIHSS (rs=-0.447, P = 0.001). The inter-rater agreement for GUSS-S was substantial (Kw=0.67, P = < 0.001). The results indicate that the GUSS-S is a valid and reliable tool for the assessment of dysphagia in acute stroke patients by different healthcare professionals.

Assessment of Home-based Monitoring in Adults with Chronic Lung Disease: An Official American Thoracic Society Research Statement

Background: There is increasing interest in the use of home-based monitoring in people with chronic lung diseases to improve access to care, support patient self-management, and facilitate the collection of information for clinical care and research. However, integration of home-based monitoring into clinical and research settings requires careful consideration of test performance and other attributes. There is no published guidance from professional respiratory societies to advance the science of home-based monitoring for chronic lung disease. Methods: An international multidisciplinary panel of 32 clinicians, researchers, patients, and caregivers developed a multidimensional framework for the evaluation of home-based monitoring in chronic lung disease developed through consensus using a modified Delphi survey. We also present an example of how the framework could be used to evaluate home-based monitoring using spirometry and pulse oximetry in adults with asthma, bronchiectasis/cystic fibrosis, chronic obstructive pulmonary disease, and interstitial lung disease. Results: The PANACEA framework includes seven domains (test Performance, disease mANAgement, Cost, patient Experience, clinician Experience, researcher Experience, and Access) to assess the degree to which home-based monitoring assessments meet the conditions for clinical and research use in chronic lung disease. Knowledge gaps and recommendations for future research of home spirometry and pulse oximetry in asthma, bronchiectasis/cystic fibrosis, chronic obstructive pulmonary disease, and interstitial lung disease were identified. Conclusions: The development of the PANACEA framework allows standardized evaluation of home-based monitoring in chronic lung diseases to support clinical application and future research.