Prediction of In-hospital Mortality Among Intensive Care Unit Patients Using Modified Daily Laboratory-based Acute Physiology Score, Version 2

BACKGROUND

Mortality prediction for intensive care unit (ICU) patients frequently relies on single ICU admission acuity measures without accounting for subsequent clinical changes.

OBJECTIVE

Evaluate novel models incorporating modified admission and daily, time-updating Laboratory-based Acute Physiology Score, version 2 (LAPS2) to predict in-hospital mortality among ICU patients.

RESEARCH DESIGN

Retrospective cohort study.

PATIENTS

ICU patients in 5 hospitals from October 2017 through September 2019.

MEASURES

We used logistic regression, penalized logistic regression, and random forest models to predict in-hospital mortality within 30 days of ICU admission using admission LAPS2 alone in patient-level and patient-day-level models, or admission and daily LAPS2 at the patient-day level. Multivariable models included patient and admission characteristics. We performed internal-external validation using 4 hospitals for training and the fifth for validation, repeating analyses for each hospital as the validation set. We assessed performance using scaled Brier scores (SBS), c -statistics, and calibration plots.

RESULTS

The cohort included 13,993 patients and 107,699 ICU days. Across validation hospitals, patient-day-level models including daily LAPS2 (SBS: 0.119-0.235; c -statistic: 0.772-0.878) consistently outperformed models with admission LAPS2 alone in patient-level (SBS: 0.109-0.175; c -statistic: 0.768-0.867) and patient-day-level (SBS: 0.064-0.153; c -statistic: 0.714-0.861) models. Across all predicted mortalities, daily models were better calibrated than models with admission LAPS2 alone.

CONCLUSIONS

Patient-day-level models incorporating daily, time-updating LAPS2 to predict mortality among an ICU population performs as well or better than models incorporating modified admission LAPS2 alone. The use of daily LAPS2 may offer an improved tool for clinical prognostication and risk adjustment in research in this population.

Health Equity in Clinical Research Informatics

OBJECTIVES

Through a scoping review, we examine in this survey what ways health equity has been promoted in clinical research informatics with patient implications and especially published in the year of 2021 (and some in 2022).

METHOD

A scoping review was conducted guided by using methods described in the Joanna Briggs Institute Manual. The review process consisted of five stages: 1) development of aim and research question, 2) literature search, 3) literature screening and selection, 4) data extraction, and 5) accumulate and report results.

RESULTS

From the 478 identified papers in 2021 on the topic of clinical research informatics with focus on health equity as a patient implication, 8 papers met our inclusion criteria. All included papers focused on artificial intelligence (AI) technology. The papers addressed health equity in clinical research informatics either through the exposure of inequity in AI-based solutions or using AI as a tool for promoting health equity in the delivery of healthcare services. While algorithmic bias poses a risk to health equity within AI-based solutions, AI has also uncovered inequity in traditional treatment and demonstrated effective complements and alternatives that promotes health equity.

CONCLUSIONS

Clinical research informatics with implications for patients still face challenges of ethical nature and clinical value. However, used prudently-for the right purpose in the right context-clinical research informatics could bring powerful tools in advancing health equity in patient care.

Predicting Survival in Patients Presenting With Spinal Epidural Metastases: The Limburg Spinal Metastasis Score

BACKGROUND

Patients with spinal epidural metastases (SEM) often experience a reduction in ambulatory status and, thus, the quality of life. Predicting which patients will benefit from a surgical intervention remains a challenge. Life expectancy is an essential factor to be considered in surgical decision-making, although not the only one. Prediction models can add value in surgical decision-making. The goal of this study was to develop and internally validate a novel model (Limburg spinal metastases score [LSMS]) and compare the predictive value with 2 commonly used models: modified Bauer score and Oswestry Spinal Risk Index (OSRI).

METHODS

We retrospectively analyzed 144 consecutive patients who underwent surgical decompression for SEM in our centers between November 2006 and December 2020. Clinical and surgical parameters were evaluated. The novel prediction model was based on multivariate analysis and was internally validated. External validation of the 2 most commonly used prediction models was performed.

RESULTS

The median survival was 17 months, 55.7% of the immobile patients regained ambulation postoperatively. In 50 patients (34.7%), at least 1 complication occurred within 30 days after surgery. The LSMS consists of 4 parameters: primary tumor type, Karnofsky performance score, presence of visceral metastases, and presence of multiple spinal metastases. Bootstrap internal validation of the model developed on this cohort yielded an optimism-corrected c-statistic of 0.75 (95% CI: 0.71-0.80). The c-statistic of the OSRI score and the Bauer score was 0.69 (95% CI: 0.64-0.74) and 0.67 (95% CI: 0.62-0.72), respectively.

CONCLUSION

The LSMS consists of 4 parameters to assist surgical decision-making for patients with SEM. The score is easy to use and appears more accurate in our population in comparison with previous existing models.

CLINICAL RELEVANCE

A novel prediction model was developed to aid in surgical decision-making for patients with spinal epidural metastases.

LEVEL OF EVIDENCE: 3

Characterizing Ultrafine Particle Mobile Monitoring Data for Epidemiology

Mobile monitoring is increasingly used to assess exposure to traffic-related air pollutants (TRAPs), including ultrafine particles (UFPs). Due to the rapid spatial decrease in the concentration of UFPs and other TRAPs with distance from roadways, mobile measurements may be non-representative of residential exposures, which are commonly used for epidemiologic studies. Our goal was to develop, apply, and test one possible approach for using mobile measurements in exposure assessment for epidemiology. We used an absolute principal component score model to adjust the contribution of on-road sources in mobile measurements to provide exposure predictions representative of cohort locations. We then compared UFP predictions at residential locations from mobile on-road plume-adjusted versus stationary measurements to understand the contribution of mobile measurements and characterize their differences. We found that predictions from mobile measurements are more representative of cohort locations after down-weighting the contribution of localized on-road plumes. Further, predictions at cohort locations derived from mobile measurements incorporate more spatial variation compared to those from short-term stationary data. Sensitivity analyses suggest that this additional spatial information captures features in the exposure surface not identified from the stationary data alone. We recommend the correction of mobile measurements to create exposure predictions representative of residential exposure for epidemiology.

Development and Internal Validation of a Prediction Model for Falls Using Electronic Health Records in a Hospital Setting

OBJECTIVE

Fall prevention is important in many hospitals. Current fall-risk-screening tools have limited predictive accuracy specifically for older inpatients. Their administration can be time-consuming. A reliable and easy-to-administer tool is desirable to identify older inpatients at higher fall risk. We aimed to develop and internally validate a prognostic prediction model for inpatient falls for older patients.

DESIGN

Retrospective analysis of a large cohort drawn from hospital electronic health record data.

SETTING AND PARTICIPANTS

Older patients (≥70 years) admitted to a university medical center (2016 until 2021).

METHODS

The outcome was an inpatient fall (≥24 hours of admission). Two prediction models were developed using regularized logistic regression in 5 imputed data sets: one model without predictors indicating missing values (Model-without) and one model with these additional predictors indicating missing values (Model-with). We internally validated our whole model development strategy using 10-fold stratified cross-validation. The models were evaluated using discrimination (area under the receiver operating characteristic curve) and calibration (plot assessment). We determined whether the areas under the receiver operating characteristic curves (AUCs) of the models were significantly different using DeLong test.

RESULTS

Our data set included 21,286 admissions. In total, 470 (2.2%) had a fall after 24 hours of admission. The Model-without had 12 predictors and Model-with 13, of which 4 were indicators of missing values. The AUCs of the Model-without and Model-with were 0.676 (95% CI 0.646-0.707) and 0.695 (95% CI 0.667-0.724). The AUCs between both models were significantly different (P = .013). Calibration was good for both models.

CONCLUSIONS AND IMPLICATIONS

Both the Model-with and Model-without indicators of missing values showed good calibration and fair discrimination, where the Model-with performed better. Our models showed competitive performance to well-established fall-risk-screening tools, and they have the advantage of being based on routinely collected data. This may substantially reduce the burden on nurses, compared with nonautomatic fall-risk-screening tools.

An Apparent Diffusion Coefficient-Based Machine Learning Model Can Improve Prostate Cancer Detection in the Grey Area of the Prostate Imaging Reporting and Data System Category 3: A Single-Centre Experience

The Prostate Imaging and Reporting Data System (PI-RADS) has a key role in the management of prostate cancer (PCa). However, the clinical interpretation of PI-RADS 3 score lesions may be challenging and misleading, thus postponing PCa diagnosis to biopsy outcome. Multiparametric magnetic resonance imaging (mpMRI) radiomic analysis may represent a stand-alone noninvasive tool for PCa diagnosis. Hence, this study aims at developing a mpMRI-based radiomic PCa diagnostic model in a cohort of PI-RADS 3 lesions. We enrolled 133 patients with 155 PI-RADS 3 lesions, 84 of which had PCa confirmation by fusion biopsy. Local radiomic features were generated from apparent diffusion coefficient maps, and the four most informative were selected using LASSO, the Wilcoxon rank-sum test (p < 0.001), and support vector machines (SVMs). The selected features where augmented and used to train an SVM classifier, externally validated on a holdout subset. Linear and second-order polynomial kernels were exploited, and their predictive performance compared through receiver operating characteristics (ROC)-related metrics. On the test set, the highest performance, equally for both kernels, was specificity = 76%, sensitivity = 78%, positive predictive value = 80%, and negative predictive value = 74%. Our findings substantially improve radiologist interpretation of PI-RADS 3 lesions and let us advance towards an image-driven PCa diagnosis.

Development and economic assessment of machine learning models to predict glycosylated hemoglobin in type 2 diabetes

Background: Glycosylated hemoglobin (HbA1c) is recommended for diagnosing and monitoring type 2 diabetes. However, the monitoring frequency in real-world applications has not yet reached the recommended frequency in the guidelines. Developing machine learning models to screen patients with poor glycemic control in patients with T2D could optimize management and decrease medical service costs. Methods: This study was carried out on patients with T2D who were examined for HbA1c at the Sichuan Provincial People's Hospital from April 2018 to December 2019. Characteristics were extracted from interviews and electronic medical records. The data (excluded FBG or included FBG) were randomly divided into a training dataset and a test dataset with a radio of 8:2 after data pre-processing. Four imputing methods, four screening methods, and six machine learning algorithms were used to optimize data and develop models. Models were compared on the basis of predictive performance metrics, especially on the model benefit (MB, a confusion matrix combined with economic burden associated with therapeutic inertia). The contributions of features were interpreted using SHapley Additive exPlanation (SHAP). Finally, we validated the sample size on the best model. Results: The study included 980 patients with T2D, of whom 513 (52.3%) were defined as positive (need to perform the HbA1c test). The results indicated that the model trained in the data (included FBG) presented better forecast performance than the models that excluded the FBG value. The best model used modified random forest as the imputation method, ElasticNet as the feature screening method, and the LightGBM algorithms and had the best performance. The MB, AUC, and AUPRC of the best model, among a total of 192 trained models, were 43475.750 (¥), 0.972, 0.944, and 0.974, respectively. The FBG values, previous HbA1c values, having a rational and reasonable diet, health status scores, type of manufacturers of metformin, interval of measurement, EQ-5D scores, occupational status, and age were the most significant contributors to the prediction model. Conclusion: We found that MB could be an indicator to evaluate the model prediction performance. The proposed model performed well in identifying patients with T2D who need to undergo the HbA1c test and could help improve individualized T2D management.

Individualized Treatment Effects of Bougie versus Stylet for Tracheal Intubation in Critical Illness

Rationale: A recent randomized trial found that using a bougie did not increase the incidence of successful intubation on first attempt in critically ill adults. The average effect of treatment in a trial population, however, may differ from effects for individuals. Objective: We hypothesized that application of a machine learning model to data from a clinical trial could estimate the effect of treatment (bougie vs. stylet) for individual patients based on their baseline characteristics ("individualized treatment effects"). Methods: This was a secondary analysis of the BOUGIE (Bougie or Stylet in Patients Undergoing Intubation Emergently) trial. A causal forest algorithm was used to model differences in outcome probabilities by randomized group assignment (bougie vs. stylet) for each patient in the first half of the trial (training cohort). This model was used to predict individualized treatment effects for each patient in the second half (validation cohort). Measurements and Main Results: Of 1,102 patients in the BOUGIE trial, 558 (50.6%) were the training cohort, and 544 (49.4%) were the validation cohort. In the validation cohort, individualized treatment effects predicted by the model significantly modified the effect of trial group assignment on the primary outcome (P value for interaction = 0.02; adjusted qini coefficient, 2.46). The most important model variables were difficult airway characteristics, body mass index, and Acute Physiology and Chronic Health Evaluation II score. Conclusions: In this hypothesis-generating secondary analysis of a randomized trial with no average treatment effect and no treatment effect in any prespecified subgroups, a causal forest machine learning algorithm identified patients who appeared to benefit from the use of a bougie over a stylet and from the use of a stylet over a bougie using complex interactions between baseline patient and operator characteristics.

Using Machine-Learning to Assess the Prognostic Value of Early Enteral Feeding Intolerance in Critically Ill Patients: A Retrospective Study

BACKGROUND

The association between gastrointestinal intolerance during early enteral nutrition (EN) and adverse clinical outcomes in critically ill patients is controversial. We aimed to assess the prognostic value of enteral feeding intolerance (EFI) markers during early ICU stays and to predict early EN failure using a machine learning (ML) approach.

METHODS

We performed a retrospective analysis of data from adult patients admitted to Beilinson Hospital ICU between January 2011 and December 2018 for more than 48 h and received EN. Clinical data, including demographics, severity scores, EFI markers, and medications, along with 72 h after admission, were analyzed by ML algorithms. Prediction performance was assessed by the area under the receiver operating characteristics (AUCROC) of a ten-fold cross-validation set.

RESULTS

The datasets comprised 1584 patients. The means of the cross-validation AUCROCs for 90-day mortality and early EN failure were 0.73 (95% CI 0.71-0.75) and 0.71 (95% CI 0.67-0.74), respectively. Gastric residual volume above 250 mL on the second day was an important component of both prediction models.

CONCLUSIONS

ML underlined the EFI markers that predict poor 90-day outcomes and early EN failure and supports early recognition of at-risk patients. Results have to be confirmed in further prospective and external validation studies.

The Problem of Pain in the United States: A Population-Based Characterization of Biopsychosocial Correlates of High Impact Chronic Pain Using the National Health Interview Survey

Over 20 million adults in the United States live with high impact chronic pain (HICP), or chronic pain that limits life or work activities for ≥3 months. It is critically important to differentiate people with HICP from those who sustain normal activities although experiencing chronic pain. Therefore, we aim to help clinicians and researchers identify those with HICP by: 1) developing models that identify factors associated with HICP using the 2016 national health interview survey (NHIS) and 2) evaluating the performances of those models overall and by sociodemographic subgroups (sex, age, and race/ethnicity). Our analysis included 32,980 respondents. We fitted logistic regression models with LASSO (a parametric model) and random forest (a nonparametric model) for predicting HICP using the whole sample. Both models performed well. The most important factors associated with HICP were those related to underlying ill-health (arthritis and rheumatism, hospitalizations, and emergency department visits) and poor psychological well-being. These factors can be used for identifying higher-risk sub-groups for screening for HICP. We will externally validate these findings in future work. We need future studies that longitudinally predict the initiation and maintenance of HICP, then use this information to prevent HICP and direct patients to optimal treatments. PERSPECTIVE: Our study developed models to identify factors associated with high-impact chronic pain (HICP) using the 2016 National Health Interview Survey. There was homogeneity in the factors associated with HICP by gender, age, and race/ethnicity. Understanding these risk factors is crucial to support the identification of populations and individuals at highest risk for developing HICP and improve access to interventions that target these high-risk subgroups.

Prediction Models for Intrauterine Growth Restriction Using Artificial Intelligence and Machine Learning: A Systematic Review and Meta-Analysis

BACKGROUND

IntraUterine Growth Restriction (IUGR) is a global public health concern and has major implications for neonatal health. The early diagnosis of this condition is crucial for obtaining positive outcomes for the newborn. In recent years Artificial intelligence (AI) and machine learning (ML) techniques are being used to identify risk factors and provide early prediction of IUGR. We performed a systematic review (SR) and meta-analysis (MA) aimed to evaluate the use and performance of AI/ML models in detecting fetuses at risk of IUGR.

METHODS

We conducted a systematic review according to the PRISMA checklist. We searched for studies in all the principal medical databases (MEDLINE, EMBASE, CINAHL, Scopus, Web of Science, and Cochrane). To assess the quality of the studies we used the JBI and CASP tools. We performed a meta-analysis of the diagnostic test accuracy, along with the calculation of the pooled principal measures.

RESULTS

We included 20 studies reporting the use of AI/ML models for the prediction of IUGR. Out of these, 10 studies were used for the quantitative meta-analysis. The most common input variable to predict IUGR was the fetal heart rate variability (n = 8, 40%), followed by the biochemical or biological markers (n = 5, 25%), DNA profiling data (n = 2, 10%), Doppler indices (n = 3, 15%), MRI data (n = 1, 5%), and physiological, clinical, or socioeconomic data (n = 1, 5%). Overall, we found that AI/ML techniques could be effective in predicting and identifying fetuses at risk for IUGR during pregnancy with the following pooled overall diagnostic performance: sensitivity = 0.84 (95% CI 0.80-0.88), specificity = 0.87 (95% CI 0.83-0.90), positive predictive value = 0.78 (95% CI 0.68-0.86), negative predictive value = 0.91 (95% CI 0.86-0.94) and diagnostic odds ratio = 30.97 (95% CI 19.34-49.59). In detail, the RF-SVM (Random Forest-Support Vector Machine) model (with 97% accuracy) showed the best results in predicting IUGR from FHR parameters derived from CTG.

CONCLUSIONS

our findings showed that AI/ML could be part of a more accurate and cost-effective screening method for IUGR and be of help in optimizing pregnancy outcomes. However, before the introduction into clinical daily practice, an appropriate algorithmic improvement and refinement is needed, and the importance of quality assessment and uniform diagnostic criteria should be further emphasized.

Predicting leptomeningeal disease spread after resection of brain metastases using machine learning

OBJECTIVE

The incidence of leptomeningeal disease (LMD) has increased as treatments for brain metastases (BMs) have improved and patients with metastatic disease are living longer. Sample sizes of individual studies investigating LMD after surgery for BMs and its risk factors have been limited, ranging from 200 to 400 patients at risk for LMD, which only allows the use of conventional biostatistics. Here, the authors used machine learning techniques to enhance LMD prediction in a cohort of surgically treated BMs.

METHODS

A conditional survival forest, a Cox proportional hazards model, an extreme gradient boosting (XGBoost) classifier, an extra trees classifier, and logistic regression were trained. A synthetic minority oversampling technique (SMOTE) was used to train the models and handle the inherent class imbalance. Patients were divided into an 80:20 training and test set. Fivefold cross-validation was used on the training set for hyperparameter optimization. Patients eligible for study inclusion were adults who had consecutively undergone neurosurgical BM treatment, had been admitted to Brigham and Women's Hospital from January 2007 through December 2019, and had a minimum of 1 month of follow-up after neurosurgical treatment.

RESULTS

A total of 1054 surgically treated BM patients were included in this analysis. LMD occurred in 168 patients (15.9%) at a median of 7.05 months after BM diagnosis. The discrimination of LMD occurrence was optimal using an XGboost algorithm (area under the curve = 0.83), and the time to LMD was prognosticated evenly by the random forest algorithm and the Cox proportional hazards model (C-index = 0.76). The most important feature for both LMD classification and regression was the BM proximity to the CSF space, followed by a cerebellar BM location. Lymph node metastasis of the primary tumor at BM diagnosis and a cerebellar BM location were the strongest risk factors for both LMD occurrence and time to LMD.

CONCLUSIONS

The outcomes of LMD patients in the BM population are predictable using SMOTE and machine learning. Lymph node metastasis of the primary tumor at BM diagnosis and a cerebellar BM location were the strongest LMD risk factors.

Prediction models of all-cause mortality among older adults in nursing home setting: A systematic review and meta-analysis

Background and Aims

Few studies have meta-analyzed different prognostic models developed for older adults, especially nursing home residents. We aimed to systematically review and meta-analyze the performance of all published models that predicted all-cause mortality among older nursing home residents.

Methods

We systematically searched PubMed and EMBASE from the databases' inception to January 1, 2020 to capture studies developing and/or validating a prognostic/prediction model for all-cause mortality among nursing home residents. We then carried out both qualitative and quantitative analyses evaluating these models' risks of bias and applicability.

Results

The systematic search yielded 23,975 articles. We identified 28 indices that predicted the risk of all-cause mortality from 14 days to 39 months among older adults in nursing homes. The most used predictors were age, sex, body weight, swallowing problem, congestive heart failure, shortness of breath, body mass index, and activities of daily living. Of the 28 indices, 8 (29%) and 3 (11%) were internally and externally validated, respectively. None of the indices was validated in more than one cohort. Of the 28 indices, 22 (79%) reported the C-statistic, while only 6 (6%) reported the 95% confidence interval for the C statistic in the development cohorts. In the validation cohorts, 11 (39%) reported the C-statistic and 8 (29%) reported the 95% confidence interval. The meta-analyzed C statistic for all indices is 0.733 (95% prediction interval: 0.669-0.797). All studies/indices had high risks of bias and high concern for applicability according to PROBAST.

Conclusion

We identified 28 indices for predicting all-cause mortality among older nursing home residents. The overall quality of evidence was low due to a high degree of bias and poor reporting of model performance statistics. Before any prediction model could be recommended in routine care, future research is needed to rigorously validate existing prediction models and evaluate their applicability and develop new prediction models.

Multi-clinical index classifier combined with AI algorithm model to predict the prognosis of gallbladder cancer

Objectives

It is significant to develop effective prognostic strategies and techniques for improving the survival rate of gallbladder carcinoma (GBC). We aim to develop the prediction model from multi-clinical indicators combined artificial intelligence (AI) algorithm for the prognosis of GBC.

Methods

A total of 122 patients with GBC from January 2015 to December 2019 were collected in this study. Based on the analysis of correlation, relative risk, receiver operator characteristic curve, and importance by AI algorithm analysis between clinical factors and recurrence and survival, the two multi-index classifiers (MIC1 and MIC2) were obtained. The two classifiers combined eight AI algorithms to model the recurrence and survival. The two models with the highest area under the curve (AUC) were selected to test the performance of prognosis prediction in the testing dataset.

Results

The MIC1 has ten indicators, and the MIC2 has nine indicators. The combination of the MIC1 classifier and the "avNNet" model can predict recurrence with an AUC of 0.944. The MIC2 classifier and "glmet" model combination can predict survival with an AUC of 0.882. The Kaplan-Meier analysis shows that MIC1 and MIC2 indicators can effectively predict the median survival of DFS and OS, and there is no statistically significant difference in the prediction results of the indicators (MIC1: χ² = 6.849, P = 0.653; MIC2: χ² = 9.14, P = 0.519).

Conclusions

The MIC1 and MIC2 combined with avNNet and mda models have high sensitivity and specificity in predicting the prognosis of GBC.

Data Quality Degradation on Prediction Models Generated From Continuous Activity and Heart Rate Monitoring: Exploratory Analysis Using Simulation

BACKGROUND

Limited data accuracy is often cited as a reason for caution in the integration of physiological data obtained from consumer-oriented wearable devices in care management pathways. The effect of decreasing accuracy on predictive models generated from these data has not been previously investigated.

OBJECTIVE

The aim of this study is to simulate the effect of data degradation on the reliability of prediction models generated from those data and thus determine the extent to which lower device accuracy might or might not limit their use in clinical settings.

METHODS

Using the Multilevel Monitoring of Activity and Sleep in Healthy People data set, which includes continuous free-living step count and heart rate data from 21 healthy volunteers, we trained a random forest model to predict cardiac competence. Model performance in 75 perturbed data sets with increasing missingness, noisiness, bias, and a combination of all 3 perturbations was compared to model performance for the unperturbed data set.

RESULTS

The unperturbed data set achieved a mean root mean square error (RMSE) of 0.079 (SD 0.001) in predicting cardiac competence index. For all types of perturbations, RMSE remained stable up to 20%-30% perturbation. Above this level, RMSE started increasing and reached the point at which the model was no longer predictive at 80% for noise, 50% for missingness, and 35% for the combination of all perturbations. Introducing systematic bias in the underlying data had no effect on RMSE.

CONCLUSIONS

In this proof-of-concept study, the performance of predictive models for cardiac competence generated from continuously acquired physiological data was relatively stable with declining quality of the source data. As such, lower accuracy of consumer-oriented wearable devices might not be an absolute contraindication for their use in clinical prediction models.

Combined Model of OCT Angiography and Structural OCT Parameters to Predict Paracentral Visual Field Loss in Primary Open-Angle Glaucoma

PURPOSE

To assess a model combining OCT angiography (OCTA) and OCT parameters to predict the severity of paracentral visual field (VF) loss in primary open-angle glaucoma (POAG).

DESIGN

Cross-sectional study.

PARTICIPANTS

Forty-four patients with POAG and 42 control subjects underwent OCTA and OCT imaging with a swept-source OCT device.

METHODS

The circumpapillary microvasculature was quantified for vessel density (cpVD) and flow (cpFlow) after delineation of Bruch's membrane opening and removal of large vessels. Retinal nerve fiber layer thickness (RNFLT) and Bruch's membrane opening-minimum rim width (BMO-MRW) were measured from structural OCT. Paracentral total deviation (PaTD) was defined as the average of the total deviation values within the central 10 degrees on Humphrey VF testing (24-2) for upper and lower hemifields. The OCT and OCTA parameters were measured in the affected hemisphere corresponding to the hemifield with lower PaTD for POAG patients. Models were created to predict affected PaTD based on RNFLT alone; RNFLT and BMO-MRW; OCTA alone; or RNFLT, BMO-MRW and OCTA parameters. The models were compared using coefficient of determination (r2) and Bayesian information criterion (BIC) score. Bayesian information criterion decrease of ≥6 indicates strong evidence for model improvement.

MAIN OUTCOME MEASURES

Performance of models containing OCT and OCTA parameters in predicting PaTD.

RESULTS

Patients with POAG and controls were similar in age and sex (65.9 ± 9.5 years and 38.4% male overall, P ≥ 0.56 for both). Average RNFLT, minimum RNFLT, average BMO-MRW, minimum BMO-MRW, cpVD, and cpFlow were all significantly lower (all P < 0.001) in the affected hemisphere in patients with POAG than in controls. In patients with POAG, the average mean deviation was -4.33 ± 3.25 dB; the PaTD of the affected hemifield averaged -4.55 ± 5.26 dB and correlated significantly with both OCTA and structural OCT parameters (r ≥ 0.43, P ≤ 0.004 for all). The model containing RNFLT, BMO-MRW, and OCTA parameters was superior in predicting affected PaTD (r2 = 0.47, BIC = 290.7), with higher r2 and lower BIC compared with all 3 other models.

CONCLUSIONS

A combined model of OCTA and structural OCT parameters can predict the severity of paracentral VF loss of the affected hemifield, supporting clinical utility of OCTA in patients with POAG with paracentral VF loss.

FINANCIAL DISCLOSURE(S)

Proprietary or commercial disclosure may be found after the references.

Validation of the ARIC prediction model for sudden cardiac death in the European population: The ESCAPE-NET project: Predicting sudden cardiac death in European adults

BACKGROUND

Sudden cardiac death is responsible for 10-20% of all deaths in Europe. The current study investigates how well the risk of sudden cardiac death can be predicted. To this end, we validated a previously developed prediction model for sudden cardiac death from the Atherosclerosis Risk in Communities study (USA).

METHODS

Data from participants of the Copenhagen City Heart Study (CCHS) (n=9988) was used to externally validate the previously developed prediction model for sudden cardiac death. The model's performance was assessed through discrimination (C-statistic) and calibration by the Hosmer-Lemeshow goodness-of-fit (HL) statistics suited for censored data and visual inspection of calibration plots. Additional validation was performed using data from the Hoorn Study (N=2045), employing the same methods.

RESULTS

During ten years of follow-up of CCHS participants (mean age: 58.7 years, 56.2% women), 425 experienced SCD (4.2%). The prediction model showed good discrimination for sudden cardiac death risk (C-statistic: 0.81, 95% CI:0.79-0.83). Calibration was robust (HL statistic: p=0.8). Visual inspection of the calibration plot showed that the calibration could be improved. Sensitivity was 89.8%, and specificity was 60.6%. The positive and negative predictive values were 10.1% and 99.2%. Model performance was similar in the Hoorn Study (C-statistic: 0.81, 95% CI: 0.77-0.85 and the HL statistic: 1.00).

CONCLUSION

Our study showed that the previously developed prediction model in North American adults performs equally well in identifying those at risk for sudden cardiac death in a general North-West European population. However, the positive predictive value is low.

Validity of the Maximal Heart Rate Prediction Models among Runners and Cyclists

Maximal heart rate (HRmax) is a widely used measure of cardiorespiratory fitness. Prediction of HRmax is an alternative to cardiopulmonary exercise testing (CPET), but its accuracy among endurance athletes (EA) requires evaluation. This study aimed to externally validate HRmax prediction models in the EA independently for running and cycling CPET. A total of 4043 runners (age = 33.6 (8.1) years; 83.5% males; BMI = 23.7 (2.5) kg·m^-2) and 1026 cyclists (age = 36.9 (9.0) years; 89.7% males; BMI = 24.0 (2.7) kg·m^-2) underwent maximum CPET. Student t-test, mean absolute percentage error (MAPE), and root mean square error (RMSE) were applied to validate eight running and five cycling HRmax equations externally. HRmax was 184.6 (9.8) beats·min^-1 and 182.7 (10.3) beats·min^-1, respectively, for running and cycling, p = 0.001. Measured and predicted HRmax differed significantly (p = 0.001) for 9 of 13 (69.2%) models. HRmax was overestimated by eight (61.5%) and underestimated by five (38.5%) formulae. Overestimated HRmax amounted to 4.9 beats·min^-1 and underestimated HRmax was in the range up to 4.9 beats·min^-1. RMSE was 9.1-10.5. MAPE ranged to 4.7%. Prediction models allow for limited precision of HRmax estimation and present inaccuracies. HRmax was more often underestimated than overestimated. Predicted HRmax can be implemented for EA as a supplemental method, but CPET is the preferable method.

Development and validation of risk prediction models for large for gestational age infants using logistic regression and two machine learning algorithms

BACKGROUND

Large for gestational age (LGA) is one of the adverse outcomes during pregnancy that endangers the life and health of mothers and offspring. We aimed to establish prediction models for LGA at late pregnancy.

METHODS

Data were obtained from an established Chinese pregnant women cohort of 1285 pregnant women. LGA was diagnosed as >90th percentile of birth weight distribution of Chinese corresponding to gestational age of the same-sex newborns. Women with gestational diabetes mellitus (GDM) were classified into three subtypes according to the indexes of insulin sensitivity and insulin secretion. Models were established by logistic regression and decision tree/random forest algorithms, and validated by the data.

RESULTS

A total of 139 newborns were diagnosed as LGA after birth. The area under the curve (AUC) for the training set is 0.760 (95% confidence interval [CI] 0.706-0.815), and 0.748 (95% CI 0.659-0.837) for the internal validation set of the logistic regression model, which consisted of eight commonly used clinical indicators (including lipid profile) and GDM subtypes. For the prediction models established by the two machine learning algorithms, which included all the variables, the training set and the internal validation set had AUCs of 0.813 (95% CI 0.786-0.839) and 0.779 (95% CI 0.735-0.824) for the decision tree model, and 0.854 (95% CI 0.831-0.877) and 0.808 (95% CI 0.766-0.850) for the random forest model.

CONCLUSION

We established and validated three LGA risk prediction models to screen out the pregnant women with high risk of LGA at the early stage of the third trimester, which showed good prediction power and could guide early prevention strategies.

Genetic and Geographical Associations With Six Dimensions of Psychotic Experiences in Adolesence

BACKGROUND AND HYPOTHESIS

Large-scale epidemiological and genetic research have shown that psychotic experiences in the community are risk factors for adverse physical and psychiatric outcomes. We investigated the associations of six types of specific psychotic experiences and negative symptoms assessed in mid-adolescence with well-established environmental and genetic risk factors for psychosis.

STUDY DESIGN

Fourteen polygenic risk scores (PRS) and nine geographical environmental variables from 3590 participants of the Twins Early Development Study (mean age 16) were associated with paranoia, hallucinations, cognitive disorganization, grandiosity, anhedonia, and negative symptoms scales. The predictors were modeled using LASSO regularization separately (Genetic and Environmental models) and jointly (GE model).

STUDY RESULTS

In joint GE models, we found significant genetic associations of negative symptoms with educational attainment PRS (β = -.07; 95% CI = -0.12 to -0.04); cognitive disorganization with neuroticism PRS (β = .05; 95% CI = 0.03-0.08); paranoia with MDD (β = .07; 95% CI = 0.04-0.1), BMI (β = .05; 95% CI = 0.02-0.08), and neuroticism PRS (β = .05; 95% CI = 0.02-0.08). From the environmental measures only family SES (β = -.07, 95% CI = -0.10 to -0.03) and regional education levels (β = -.06; 95% CI = -0.09 to -0.02) were associated with negative symptoms.

CONCLUSIONS

Our findings advance understanding of how genetic propensity for psychiatric, cognitive, and anthropometric traits, as well as environmental factors, together play a role in creating vulnerability for specific psychotic experiences and negative symptoms in mid-adolescence.

Proof-of-concept Study of Using Supervised Machine Learning Algorithms to Predict Self-care and Glycemic Control in Type 1 Diabetes Patients on Insulin Pump Therapy

OBJECTIVE

Using supervised machine learning algorithms (SMLAs), we built models to predict the probability of type 1 diabetes mellitus (T1DM) patients on insulin pump therapy for meeting insulin pump self-management behavioral (IPSMB) criteria and achieving good glycemic response within six months.

METHODS

This was a single-center retrospective chart review of 100 adult T1DM patients on insulin pump therapy (>6 months). Three SMLAs were deployed: multivariable logistic regression (LR), random forest (RF), and K-nearest neighbor (k-NN); validated using repeated three-fold cross-validation. Performance metrics included AUC-ROC for discrimination and Brier scores for calibration.

RESULTS

Variables predictive of adherence with IPSMB criteria were baseline HbA1c, continuous glucose monitoring (CGM), and sex. The models had comparable discriminatory power (LR=0.74; RF=0.74; k-NN=0.72), with the random forest model showing better calibration (Brier=0.151). Predictors of the good glycemic response included baseline HbA1c, entering carbohydrates, and following the recommended bolus dose, with models comparable in discriminatory power (LR=0.81, RF=0.80, k-NN=0.78) but the random forest model being better calibrated (Brier=0.099).

CONCLUSION

These proof-of-concept analyses demonstrate the feasibility of using SMLAs to develop clinically relevant predictive models of adherence with IPSMB criteria and glycemic control within six months. Subject to further study, non-linear prediction models may perform better.

Electronic health record-based prediction models for in-hospital adverse drug event diagnosis or prognosis: a systematic review

OBJECTIVE

We conducted a systematic review to characterize and critically appraise developed prediction models based on structured electronic health record (EHR) data for adverse drug event (ADE) diagnosis and prognosis in adult hospitalized patients.

MATERIALS AND METHODS

We searched the Embase and Medline databases (from January 1, 1999, to July 4, 2022) for articles utilizing structured EHR data to develop ADE prediction models for adult inpatients. For our systematic evidence synthesis and critical appraisal, we applied the Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modelling Studies (CHARMS).

RESULTS

Twenty-five articles were included. Studies often did not report crucial information such as patient characteristics or the method for handling missing data. In addition, studies frequently applied inappropriate methods, such as univariable screening for predictor selection. Furthermore, the majority of the studies utilized ADE labels that only described an adverse symptom while not assessing causality or utilizing a causal model. None of the models were externally validated.

CONCLUSIONS

Several challenges should be addressed before the models can be widely implemented, including the adherence to reporting standards and the adoption of best practice methods for model development and validation. In addition, we propose a reorientation of the ADE prediction modeling domain to include causality as a fundamental challenge that needs to be addressed in future studies, either through acquiring ADE labels via formal causality assessments or the usage of adverse event labels in combination with causal prediction modeling.

An In-Hospital Mortality Risk Model for Elderly Patients Undergoing Cardiac Valvular Surgery Based on LASSO-Logistic Regression and Machine Learning

BACKGROUND

To preferably evaluate and predict the risk for in-hospital mortality in elderly patients receiving cardiac valvular surgery, we developed a new prediction model using least absolute shrinkage and selection operator (LASSO)-logistic regression and machine learning (ML) algorithms.

METHODS

Clinical data including baseline characteristics and peri-operative data of 7163 elderly patients undergoing cardiac valvular surgery from January 2016 to December 2018 were collected at 87 hospitals in the Chinese Cardiac Surgery Registry (CCSR). Patients were divided into training (N = 5774 [80%]) and testing samples (N = 1389 [20%]) according to their date of operation. LASSO-logistic regression models and ML models were used to analyze risk factors and develop the prediction model. We compared the discrimination and calibration of each model and EuroSCORE II.

RESULTS

A total of 7163 patients were included in this study, with a mean age of 69.8 (SD 4.5) years, and 45.0% were women. Overall, in-hospital mortality was 4.05%. The final model included seven risk factors: age, prior cardiac surgery, cardiopulmonary bypass duration time (CPB time), left ventricular ejection fraction (LVEF), creatinine clearance rate (CCr), combined coronary artery bypass grafting (CABG) and New York Heart Association (NYHA) class. LASSO-logistic regression, linear discriminant analysis (LDA), support vector classification (SVC) and logistic regression (LR) models had the best discrimination and calibration in both training and testing cohorts, which were superior to the EuroSCORE II.

CONCLUSIONS

The mortality rate for elderly patients undergoing cardiac valvular surgery was relatively high. LASSO-logistic regression, LDA, SVC and LR can predict the risk for in-hospital mortality in elderly patients receiving cardiac valvular surgery well.

Empirical comparison of routinely collected electronic health record data for head and neck cancer-specific survival in machine-learnt prognostic models

BACKGROUND

Knowledge of the prognostic factors and performance of machine learning predictive models for 2-year cancer-specific survival (CSS) is limited in the head and neck cancer (HNC) population.

METHODS

Data from our facilities' oncology information system (OIS) collected for routine practice (OIS dataset, n = 430 patients) and research purposes (research dataset, n = 529 patients) were extracted on adults diagnosed between 2000 and 2017 with squamous cell carcinoma of the head and neck.

RESULTS

Machine learning demonstrated excellent performance (area under the curve, AUC) in the whole cohort (AUC = 0.97, research dataset), larynx cohort (AUC = 0.98, both datasets), and oropharynx cohort (AUC = 0.99, both datasets). Tumor site and T classification were identified as predictors of 2-year CSS in both datasets. Hypothyroidism and fitness for operation were further identified in the research dataset.

CONCLUSIONS

Datasets extracted from an OIS for routine clinical practice and research purposes demonstrated high utility for informing 2-year head and neck CSS.

An Analytical Framework on Utilizing Various Integrated Multi-Trophic Scenarios for Basil Production

Here, we aim to improve the overall sustainability of aquaponic basil (Ocimum basilicum L.)-sturgeon (Acipenser baerii) integrated recirculating systems. We implement new AI methods for operational management together with innovative solutions for plant growth bed, consisting of Rapana venosa shells (R), considered wastes in the food processing industry. To this end, the ARIMA-supervised learning method was used to develop solutions for forecasting the growth of both fish and plant biomass, while multi-linear regression (MLR), generalized additive models (GAM), and XGBoost were used for developing black-box virtual sensors for water quality. The efficiency of the new R substrate was evaluated and compared to the consecrated light expended clay aggregate-LECA aquaponics substrate (H). Considering two different technological scenarios (A-high feed input, B-low feed input, respectively), nutrient reduction rates, plant biomass growth performance and additionally plant quality are analysed. The resulting prediction models reveal a good accuracy, with the best metrics for predicting N-NO₃ concentration in technological water. Furthermore, PCA analysis reveals a high correlation between water dissolved oxygen and pH. The use of innovative R growth substrate assured better basil growth performance. Indeed, this was in terms of both average fresh weight per basil plant, with 22.59% more at AR compared to AH, 16.45% more at BR compared to BH, respectively, as well as for average leaf area (LA) with 8.36% more at AR compared to AH, 9.49% more at BR compared to BH. However, the use of R substrate revealed a lower N-NH₄ and N-NO₃ reduction rate in technological water, compared to H-based variants (19.58% at AR and 18.95% at BR, compared to 20.75% at AH and 26.53% at BH for N-NH₄; 2.02% at AR and 4.1% at BR, compared to 3.16% at AH and 5.24% at BH for N-NO₃). The concentration of Ca, K, Mg and NO₃ in the basil leaf area registered the following relationship between the experimental variants: AR > AH > BR > BH. In the root area however, the NO₃ were higher in H variants with low feed input. The total phenolic and flavonoid contents in basil roots and aerial parts and the antioxidant activity of the methanolic extracts of experimental variants revealed that the highest total phenolic and flavonoid contents were found in the BH variant (0.348% and 0.169%, respectively in the roots, 0.512% and 0.019%, respectively in the aerial parts), while the methanolic extract obtained from the roots of the same variant showed the most potent antioxidant activity (89.15%). The results revealed that an analytical framework based on supervised learning can be successfully employed in various technological scenarios to optimize operational management in an aquaponic basil (Ocimum basilicum L.)-sturgeon (Acipenser baerii) integrated recirculating systems. Also, the R substrate represents a suitable alternative for replacing conventional aquaponic grow beds. This is because it offers better plant growth performance and plant quality, together with a comparable nitrogen compound reduction rate. Future studies should investigate the long-term efficiency of innovative R aquaponic growth bed. Thus, focusing on the application of the developed prediction and forecasting models developed here, on a wider range of technological scenarios.