Statistical evaluation of prognostic versus diagnostic models: beyond the ROC curve

Development and validation of a prognosis risk score model for neonatal mortality in the Amhara region, Ethiopia. A prospective cohort study

BACKGROUND

A neonatal mortality prediction score can assist clinicians in making timely clinical decisions to save neonates' lives by facilitating earlier admissions where needed. It can also help reduce unnecessary admissions.

OBJECTIVE

The study aimed to develop and validate a prognosis risk score for neonatal mortality within 28 days in public hospitals in the Amhara region, Ethiopia.

METHODS

The model was developed using a validated neonatal near miss assessment scale and a prospective cohort of 365 near-miss neonates in six hospitals between July 2021 and January 2022. The model's accuracy was assessed using the area under the receiver operating characteristics curve, calibration belt, and the optimism statistic. Internal validation was performed using a 500-repeat bootstrapping technique. Decision curve analysis was used to evaluate the model's clinical utility.

RESULTS

In total, 63 of the 365 neonates died, giving a neonatal mortality rate of 17.3% (95% CI: 13.7-21.5). Six potential predictors were identified and included in the model: anemia during pregnancy, pregnancy-induced hypertension, gestational age less than 37 weeks, birth asphyxia, 5 min Apgar score less than 7, and birth weight less than 2500 g. The model's AUC was 84.5% (95% CI: 78.8-90.2). The model's predictive ability while accounting for overfitting via internal validity was 82%. The decision curve analysis showed higher clinical utility performance.

CONCLUSION

The neonatal mortality predictive score could aid in early detection, clinical decision-making, and, most importantly, timely interventions for high-risk neonates, ultimately saving lives in Ethiopia.

External Validation of the Spinal Orthopedic Research Group Index for Spinal Epidural Abscess 90-Day Mortality in a Geographically Remote Population

STUDY DESIGN

Retrospective cohort study.

OBJECTIVE

To externally validate the Spinal Orthopaedic Research Group (SORG) index for predicting 90-day mortality from spinal epidural abscess and compare its utility to the 11-item modified frailty index (mFI-11) and Charlson comorbidity index (CCI).

SUMMARY OF BACKGROUND DATA

Providing a mortality estimate may guide informed patient and clinician decision-making. A number of prognostic tools and calculators are available to help predict the risk of mortality from spinal epidural abscess, including the SORG index, which estimates 90-day postdischarge mortality. External validation is essential before wider use of any clinical prediction tool.

MATERIALS AND METHODS

Patients were identified using hospital coding. Medical and radiologic records were used to confirm the diagnosis. Mortality data and data to calculate the SORG index, mFI-11, and CCI were collected. Area under the curve and calibration plots were used to analyze.

RESULTS

One hundred and fifty patients were included: 58 were female (39%), with a median age of 63 years. Fifteen deaths (10%) at 90 days postdischarge and 20 (13%) at one year. The mean SORG index was 13.6%, the mean CCI 2.75, and the mean mFI-11 was 1.34. The SORG index ( P =0.0006) and mFI-11 ( P <0.0001) were associated with 90-day mortality. Area under the curve for SORG, mFI-11, and CCI were 0.81, 0.84, and 0.49, respectively. The calibration slope for the SORG index showed slight overestimation in the middle ranges of the predicted probability, more so than mFI-11, and was not well-calibrated over the higher ranges of predicted probability.

CONCLUSIONS

This study externally validated the SORG index, demonstrating its utility in our population at predicting 90-day mortality; however, it was less well calibrated than the mFI-11. Variations in algorithm performance may be a result of differences in socioethnic composition and health resources between development and validation centres. Continued multicenter data input may help improve such algorithms and their generalisability.

Development and validation of a prediction model for people with mild chronic kidney disease in Japanese individuals

BACKGROUND

Chronic kidney disease (CKD) poses significant health risks due to its asymptomatic nature in early stages and its association with increased cardiovascular and kidney events. Early detection and management are critical for improving outcomes.

OBJECTIVE

This study aimed to develop and validate a prediction model for hospitalization for ischemic heart disease (IHD) or cerebrovascular disease (CVD) and major kidney events in Japanese individuals with mild CKD using readily available health check and prescription data.

METHODS

A retrospective cohort study was conducted using data from approximately 850,000 individuals in the PREVENT Inc. database, collected between April 2013 and April 2023. Cox proportional hazard regression models were utilized to derive and validate risk scores for hospitalization for IHD/CVD and major kidney events, incorporating traditional risk factors and CKD-specific variables. Model performance was assessed using the concordance index (c-index) and 5-fold cross-validation.

RESULTS

A total of 40,351 individuals were included. Key predictors included age, sex, diabetes, hypertension, and lipid levels for hospitalization for IHD/CVD and major kidney events. Age significantly increased the risk score for both hospitalization for IHD/CVD and major kidney events. The baseline 5-year survival rates are 0.99 for hospitalization for IHD/CVD and major kidney events are 0.99. The developed risk models demonstrated predictive ability, with mean c-indexes of 0.75 for hospitalization for IHD/CVD and 0.69 for major kidney events.

CONCLUSIONS

This prediction model offers a practical tool for early identification of Japanese individuals with mild CKD at risk for hospitalization for IHD/CVD and major kidney events, facilitating timely interventions to improve patient outcomes and reduce healthcare costs. The models stratified patients into risk categories, enabling identification of those at higher risk for adverse events. Further clinical validation is required.

Urine Output and Acute Kidney Injury in Neonates/Younger Children: A Prospective Study of Cardiac Surgery Patients with Indwelling Urinary Catheters

Key Points

Using indwelling urinary catheters, urine output (UO) shows good performance in neonates and younger children. Using higher UO thresholds in neonates post-cardiac surgery improves discriminatory capacity for outcomes compared to neonatal Kidney Disease Improving Global Outcomes. In younger children (1–24 months), higher UO thresholds were not better than the adult Kidney Disease Improving Global Outcomes criteria.

Background

Pediatric AKI is associated with significant morbidity and mortality, yet a precise definition, especially concerning urine output (UO) thresholds, remains unproven. We evaluate UO thresholds for AKI in neonates and children aged 1–24 months with indwelling urinary catheters undergoing cardiac surgery.

Methods

A 6-year prospective cohort study (2018–2023) after cardiac surgery was conducted at a reference center in Brazil. All patients had indwelling urinary catheters up to 48 hours after surgery and at least two serum creatinine measurements, including one before surgery. The main objective of this study was to determine the optimal UO thresholds for AKI definition and staging in neonates and younger children compared with the currently used criteria—neonatal and adult Kidney Disease Improving Global Outcomes (KDIGO) definitions. The outcome was a composite of severe AKI (stage 3 AKI diagnosed by the serum creatinine criterion only), KRT, or hospital mortality.

Results

The study included 1024 patients: 253 in the neonatal group and 772 in the younger children group. In both groups, the lowest UO at 24 hours as a continuous variable had good discriminatory capacity for the composite outcome (area under the curve-receiver operating characteristic 0.75 [95% confidence interval, 0.70 to 0.81] and 0.74 [95% confidence interval, 0.68 to 0.79]). In neonates, the best thresholds were 3.0, 2.0, and 1.0 ml/kg per hour, and in younger children, the thresholds were 1.8, 1.0, and 0.5 ml/kg per hour. These values were used for modified AKI staging for each age group. In neonates, this modified criterion was associated with the best discriminatory capacity (area under the curve-receiver operating characteristic 0.74 [0.67 to 0.80] versus 0.68 [0.61 to 0.75], P < 0.05) and net reclassification improvement in comparison with the neonatal KDIGO criteria. In younger children, the modified criteria had good discriminatory capacity but were comparable with the adult KDIGO criteria, and the net reclassification improvement was near zero.

Conclusions

Using indwelling catheters for UO measurements, our study reinforced that the current KDIGO criteria may require adjustments to better serve the neonate population. In addition, using the UO criteria, we validated the adult KDIGO criteria in children aged 1–24 months.

External validation of risk prediction model for gestational diabetes: Individual participant data meta-analysis of randomized trials

BACKGROUND

An original validated risk prediction model with good discriminatory prognostic performance for predicting gestational diabetes (GDM) diagnosis, has been updated for recent international association of diabetes in pregnancy study group (IADPSG) diagnostic criteria. However, the updated model is yet to be externally validated on an international dataset.

AIMS

To perform an external validation of the updated risk prediction model to evaluate model indices such as discrimination and calibration based on data from the International Weight Management in Pregnancy (i-WIP) Collaborative Group.

MATERIALS AND METHODS

The i -WIP dataset was used to validate the GDM prediction tool across discrimination and model calibration.

RESULTS

Overall 7689 individual patient data were included, with 17.4 % with GDM, however only 113 cases were available using IADPSG (International Association of Diabetes and Pregnancy Groups) criteria for 75 g OGTT glucose load and ACOG (American College of Obstetricians and Gynecologists) for 100 g glucose load and having the routine clinical risk factor data. The GDM model was moderately discriminatory (Area Under the Curve (AUC) of 0.67; 95 % CI 0.59 to 0.75), Sensitivity 81.0 % (95 % CI 66.7 % to 90.9 %), specificity 53 % (40.3 % to 65.4 %). The GDM score showed reasonable calibration for predicting GDM (slope = 0.84, CITL = 0.77). Imputation for missing data increased the sample to n = 253, and vastly improved the discrimination and calibration of the model to AUC = 78 (95 % CI 72 to 85), sensitivity (81 %, 95 % CI 66.7 % to 90.9 %) and specificity (75 %, 95 % CI 68.8 % to 81 %).

CONCLUSION

The updated GDM model showed promising discrimination in predicting GDM in an international population sourced from RCT individual patient data. External validations are essential in order for the risk prediction area to advance, and we demonstrate the utility of using existing RCT data from different global settings. Despite limitations associated with harmonising the data to the variable types in the model, the validation model indices were reasonable, supporting generalizability across continents and populations.

LOCC: a novel visualization and scoring of cutoffs for continuous variables with hepatocellular carcinoma prognosis as an example

BACKGROUND

The interpretation of large datasets, such as The Cancer Genome Atlas (TCGA), for scientific and research purposes, remains challenging despite their public availability. In this study, we focused on identifying gene expression profiles most relevant to patient prognosis and aimed to develop a method and database to address this issue. To achieve this, we introduced Luo's Optimization Categorization Curve (LOCC), an innovative tool for visualizing and scoring continuous variables against dichotomous outcomes. To demonstrate the efficacy of LOCC using real-world data, we analyzed gene expression profiles and patient data from TCGA hepatocellular carcinoma samples.

RESULTS

To showcase LOCC, we demonstrate an optimal cutoff for E2F1 expression in hepatocellular carcinoma, which was subsequently validated in an independent cohort. Compared to ROC curves and their AUC, LOCC offered a superior description of the predictive value of E2F1 expression across various cancer types. The LOCC score, comprised of factors representing significance, range, and impact of the biomarker, facilitated the ranking of all gene expression profiles in hepatocellular carcinoma, aiding in the evaluation and understanding of previously published prognostic gene signatures. We also demonstrate that LOCC does not have the same assumptions required of Cox proportional hazards modeling for accurate analysis. Repeated sampling demonstrated that LOCC scores outperformed ROC's AUC in discriminating predictors from non-predictors. Additionally, gene set enrichment analysis revealed significant associations between certain genes and prognosis, such as E2F target genes and G2M checkpoint with poor prognosis, and bile acid metabolism and oxidative phosphorylation with good prognosis.

CONCLUSION

In summary, we present LOCC as a novel visualization tool for the analysis of gene expression in cancer, particularly for understanding and selecting cutoffs. Our findings suggest that LOCC scores, which effectively rank genes based on their prognostic potential, represent a more suitable approach than ROC curves and Cox proportional hazard for prognostic modeling and understanding in cancer gene expression analysis. LOCC holds promise as an invaluable tool for advancing precision medicine and furthering biomarker research. Further research regarding multivariable integration and validation will help LOCC reach its full potential and establish its utility across diverse cancer types and clinical settings.

A plasma proteomic signature for atherosclerotic cardiovascular disease risk prediction in the UK Biobank cohort

Background

While risk stratification for atherosclerotic cardiovascular disease (ASCVD) is essential for primary prevention, current clinical risk algorithms demonstrate variability and leave room for further improvement. The plasma proteome holds promise as a future diagnostic and prognostic tool that can accurately reflect complex human traits and disease processes. We assessed the ability of plasma proteins to predict ASCVD.

Method

Clinical, genetic, and high-throughput plasma proteomic data were analyzed for association with ASCVD in a cohort of 41,650 UK Biobank participants. Selected features for analysis included clinical variables such as a UK-based cardiovascular clinical risk score (QRISK3) and lipid levels, 36 polygenic risk scores (PRSs), and Olink protein expression data of 2,920 proteins. We used least absolute shrinkage and selection operator (LASSO) regression to select features and compared area under the curve (AUC) statistics between data types. Randomized LASSO regression with a stability selection algorithm identified a smaller set of more robustly associated proteins. The benefit of plasma proteins over standard clinical variables, the QRISK3 score, and PRSs was evaluated through the derivation of Δ AUC values. We also assessed the incremental gain in model performance using proteomic datasets with varying numbers of proteins. To identify potential causal proteins for ASCVD, we conducted a two-sample Mendelian randomization (MR) analysis.

Result

The mean age of our cohort was 56.0 years, 60.3% were female, and 9.8% developed incident ASCVD over a median follow-up of 6.9 years. A protein-only LASSO model selected 294 proteins and returned an AUC of 0.723 (95% CI 0.708-0.737). A clinical variable and PRS-only LASSO model selected 4 clinical variables and 20 PRSs and achieved an AUC of 0.726 (95% CI 0.712-0.741). The addition of the full proteomic dataset to clinical variables and PRSs resulted in a Δ AUC of 0.010 (95% CI 0.003-0.018). Fifteen proteins selected by a stability selection algorithm offered improvement in ASCVD prediction over the QRISK3 risk score [Δ AUC: 0.013 (95% CI 0.005-0.021)]. Filtered and clustered versions of the full proteomic dataset (consisting of 600-1,500 proteins) performed comparably to the full dataset for ASCVD prediction. Using MR, we identified 11 proteins as potentially causal for ASCVD.

Conclusion

A plasma proteomic signature performs well for incident ASCVD prediction but only modestly improves prediction over clinical and genetic factors. Further studies are warranted to better elucidate the clinical utility of this signature in predicting the risk of ASCVD over the standard practice of using the QRISK3 score.

Hybridizing mechanistic modeling and deep learning for personalized survival prediction after immune checkpoint inhibitor immunotherapy

We present a study where predictive mechanistic modeling is combined with deep learning methods to predict individual patient survival probabilities under immune checkpoint inhibitor (ICI) immunotherapy. This hybrid approach enables prediction based on both measures that are calculable from mechanistic models of key mechanisms underlying ICI therapy that may not be directly measurable in the clinic and easily measurable quantities or patient characteristics that are not always readily incorporated into predictive mechanistic models. A deep learning time-to-event predictive model trained on a hybrid mechanistic + clinical data set from 93 patients achieved higher per-patient predictive accuracy based on event-time concordance, Brier score, and negative binomial log-likelihood-based criteria than when trained on only mechanistic model-derived values or only clinical data. Feature importance analysis revealed that both clinical and model-derived parameters play prominent roles in increasing prediction accuracy, further supporting the advantage of our hybrid approach.

Mass Spectrometry Imaging of the Subchondral Bone in Osteoarthritis Reveals Tissue Remodeling of Extracellular Matrix Proteins that Precede Cartilage Loss

Osteoarthritis (OA) of the knee is a degenerative condition of the skeletal extracellular matrix (ECM) marked by the loss of articular cartilage and subchondral bone homeostasis. Treatments for OA in the knee beyond full joint replacement are lacking primarily due to gaps in molecular knowledge of the biological drivers of disease. Here, Mass Spectrometry Imaging (MSI) enabled molecular spatial mapping of the proteomic landscape of human knee tissues. Histologic sections of human tibial plateaus from OA patients and cadaveric controls were treated with collagenase III to target ECM proteins prior to imaging using a timsTOF fleX mass spectrometer (Bruker) for matrix-assisted laser desorption ionization (MALDI)-MSI of bone and cartilage proteins in human knees. Spatial MSI data of the knee, using sections of the tibial plateau from non-arthritic, cadaveric donors or from knee replacement patients with medial OA were processed and automatically segmented identifying distinct areas of joint damage. ECM peptide markers compared either OA to cadaveric tissues or OA medial to OA lateral. Not only did candidate peptides distinguish OA relative to intact cartilage, but also emphasized a significant spatial difference between OA and intact subchondral bone (AUROC >0.85). Overall, 31 peptide candidates from ECM proteins, including COL1A1, COL3A1, and unanticipated detection of collagens COL6A1 and COL6A3 in adult bone, exhibited significantly elevated abundance in diseased tissue. Highly specific hydroxyproline-containing collagens dominated OA subchondral bone directly under regions of lost cartilage revealing dramatic tissue remodeling providing molecular details on the progression of joint degeneration in OA. The identification of specific spatial markers for the progression of subchondral bone degeneration in OA advances our molecular understanding of coupled deterioration of joint tissues.

ROC curve analysis: a useful statistic multi-tool in the research of nephrology

In the past decade, scientific research in the area of Nephrology has focused on evaluating the clinical utility and performance of various biomarkers for diagnosis, risk stratification and prognosis. Before implementing a biomarker in everyday clinical practice for screening a specific disease context, specific statistic measures are necessary to evaluate the diagnostic accuracy and performance of this biomarker. Receiver Operating Characteristic (ROC) Curve analysis is an important statistical method used to estimate the discriminatory performance of a novel diagnostic test, identify the optimal cut-off value for a test that maximizes sensitivity and specificity, and evaluate the predictive value of a certain biomarker or risk, prediction score. Herein, through practical examples, we aim to present a simple methodological approach to explain in detail the principles and applications of ROC curve analysis in the field of nephrology pertaining diagnosis and prognosis.

Multivariate Analysis of Factors Associated with Feeding Mother's Own Milk at Discharge in Preterm Infants: A Retrospective Cohort Study

OBJECTIVE

 This study aimed to develop a predictive model of feeding mother's own milk (MOM) at discharge using social determinants of health (SDOH), maternal and neonatal factors after deliveries at <33 weeks of gestational age (GA), or birth weight <1,500 g.

STUDY DESIGN

 Secondary analysis of a retrospective cohort in an inner-city hospital before (Epoch-1, 2018-2019) and after (Epoch-2, 2020-2021) implementing a donor human milk (DHM) program.

RESULTS

 Among 986 neonates, 495 were born in Epoch-1 (320 Hispanic White, 142 Non-Hispanic Black, and 33 Other) and 491 in Epoch-2 (327, 137, and 27, respectively). Feeding any MOM was less frequent in infants of non-Hispanic Black mothers than in those of Hispanic mothers (p < 0.05) but did not change with epoch (p = 0.46). Among infants who received any MOM, continued feeding MOM to the time of discharge was less frequent in infants of non-Hispanic Black mothers versus those of Hispanic mothers, 94/237 (40%) versus 339/595 (57%; p < 0.05), respectively. In multivariate analysis including SDOH and maternal variables, the odds of feeding MOM at discharge were lower with SDOH including neighborhoods with higher poverty levels, multiparity, substance use disorder, non-Hispanic Black versus Hispanic and young maternal age and increased with GA but did not change after implementing DHM. The predictive model including SDOH, maternal and early neonatal variables had good discrimination (area under the curve 0.85) and calibration and was internally validated. It showed the odds of feeding MOM at discharge were lower in infants of non-Hispanic Black mothers and with feeding DHM, higher need for respiratory support and later initiation of feeding MOM.

CONCLUSION

 Feeding MOM at discharge was associated with SDOH, and maternal and neonatal factors but did not change after implementing DHM. Disparity in feeding MOM at discharge was explained by less frequent initiation and shorter duration of feeding MOM but not by later initiation of feeding MOM.

KEY POINTS

· In this cohort study of preterm infants, factors of feeding MOM at discharge included (1) SDOH; (2) postnatal age at initiation of feeding MOM; and (3) maternal and neonatal factors.. · Feeding MOM at the time of discharge was less frequent in infants of non-Hispanic Black mothers versus those of Hispanic mothers.. · Disparity in feeding MOM at discharge was explained by less frequent initiation and shorter duration of MOM feeding but not by later postnatal age at initiation of feeding MOM..

Machine Learning for Predicting Chronic Renal Disease Progression in COVID-19 Patients with Acute Renal Injury: A Feasibility Study

This study aimed to determine the feasibility of applying machine-learning methods to assess the progression of chronic kidney disease (CKD) in patients with coronavirus disease (COVID-19) and acute renal injury (AKI). The study was conducted on patients aged 18 years or older who were diagnosed with COVID-19 and AKI between April 2020 and March 2021, and admitted to a second-level hospital in Mérida, Yucatán, México. Of the admitted patients, 47.92% died and 52.06% were discharged. Among the discharged patients, 176 developed AKI during hospitalization, and 131 agreed to participate in the study. The study's results indicated that the area under the receiver operating characteristic curve (AUC-ROC) for the four models was 0.826 for the support vector machine (SVM), 0.828 for the random forest, 0.840 for the logistic regression, and 0.841 for the boosting model. Variable selection methods were utilized to enhance the performance of the classifier, with the SVM model demonstrating the best overall performance, achieving a classification rate of 99.8% ± 0.1 in the training set and 98.43% ± 1.79 in the validation set in AUC-ROC values. These findings have the potential to aid in the early detection and management of CKD, a complication of AKI resulting from COVID-19. Further research is required to confirm these results.

Development of antepartum risk prediction model for postpartum hemorrhage in Lagos, Nigeria: A prospective cohort study (Predict-PPH study)

OBJECTIVES

There is currently a limited ability to accurately identify women at risk of postpartum hemorrhage (PPH). We conducted the "Predict-PPH" study to develop and evaluate an antepartum prediction model and its derived risk-scoring system.

METHODS

This was a prospective cohort study of healthy pregnant women who registered and gave birth in five hospitals in Lagos, Nigeria, from January to June 2023. Maternal antepartum characteristics were compared between women with and without PPH. A predictive multivariable model was estimated using binary logistic regression with a backward stepwise approach eliminating variables when P was greater than 0.10. Statistically significant associations in the final model were reported when P was less than 0.05.

RESULTS

The prevalence of PPH in the enrolled cohort was 37.1%. Independent predictors of PPH such as maternal obesity (adjusted odds ratio [aOR] 3.25, 95% confidence interval [CI] 2.47-4.26), maternal anemia (aOR 1.32, 95% CI 1.02-1.72), previous history of cesarean delivery (aOR 4.24, 95% CI 3.13-5.73), and previous PPH (aOR 2.65, 95% CI 1.07-6.56) were incorporated to develop a risk-scoring system. The area under the receiver operating characteristic curve (AUROC) for the prediction model and risk scoring system was 0.72 (95% CI 0.69-0.75).

CONCLUSION

We recorded a relatively high prevalence of PPH. Our model performance was satisfactory in identifying women at risk of PPH. Therefore, the derived risk-scoring system could be a useful tool to screen and identify pregnant women at risk of PPH during their routine antenatal assessment for birth preparedness and complication readiness.

A competing risk predictive model for kidney failure in patients with advanced chronic kidney disease

BACKGROUND/PURPOSE

Predictive modeling aids in identifying patients at high risk of adverse events. Using routinely collected data, we report a competing risk prediction model for kidney failure.

METHODS

A total of 5138 patients with CKD stages 3b-5 were included and randomized into the development and validation cohorts at a ratio of 7:3. The outcome was end-stage kidney disease, defined as the initiation of dialysis or kidney transplantation. All patients were followed-up until December 31, 2020. A Fine and Gray model was applied to estimate the sub-hazard ratio of kidney failure, with death as a competing event.

RESULTS

In the development cohort, the mean age was 67.6 ± 13.9 years and 60 % were male. The mean index eGFR and median urinary protein-creatinine ratio (UPCR) were 26.5 ± 12.8 mL/min/1.73 m2 and 1051 mg/g, respectively. The median follow-up duration was 1051 days. The proportion of patients with kidney failure and death was 25.4 % and 14.1 %, respectively. Four models were applied, including eGFR, age, sex, UPCR, systolic and diastolic blood pressure, serum albumin, phosphate, uric acid, haemoglobin, and potassium levels had the best goodness of fit. All models had good discrimination with time-to-event c statistics of 0.89-0.95 in the development cohort and 0.86-0.95 in the validation cohort. The prediction models showed excellent and fairly good calibration at 2 and 5-year risk, respectively.

CONCLUSION

Using real-world data, our competing risk model can accurately predict progression to kidney failure over 2 years in patients with advanced CKD.

The role of serum magnesium in the prediction of acute kidney injury after total aortic arch replacement: A prospective observational study

Background

Considerable morbidity and death are associated with acute kidney damage (AKI) following total aortic arch replacement (TAAR). The relationship between AKI following TAAR and serum magnesium levels remains unknown. The intention of this research was to access the predictive value of serum magnesium levels on admission to the Cardiovascular Surgical Intensive Care Unit (CSICU) for AKI in patients receiving TAAR.

Methods

From May 2018 to January 2020, a prospective, observational study was performed in the Guangdong Provincial People's Hospital CSICU. Patients accepting TAAR admitted to the CSICU were studied. The Kidney Disease: Improving Global Outcomes (KDIGO) definition of serum creatinine was used to define AKI, and KDIGO stages two or three were used to characterize severe AKI. Multivariable logistic regression and area under the curve receiver-operator characteristic curve (AUC-ROC) analysis were conducted to assess the predictive capability of the serum magnesium for AKI detection. Finally, the prediction model for AKI was established and internally validated.

Results

Of the 396 enrolled patients, AKI occurred in 315 (79.5%) patients, including 154 (38.8%) patients with severe AKI. Serum magnesium levels were independently related to the postoperative AKI and severe AKI (both, P < 0.001), and AUC-ROCs for predicting AKI and severe AKI were 0.707 and 0.695, respectively. Across increasing quartiles of serum magnesium, the multivariable-adjusted odds ratios (95% confidence intervals) of postoperative AKI were 1.00 (reference), 1.04 (0.50-2.82), 1.20 (0.56-2.56), and 6.19 (2.02-23.91) (P for Trend < 0.001). When serum magnesium was included to a baseline model with established risk factors, AUC-ROC (0.833 vs 0.808, P = 0.050), reclassification (P < 0.001), and discrimination (P = 0.002) were further improved.

Conclusions

Serum magnesium levels on admission are an independent predictor of AKI. In TAAR patients, elevated serum magnesium levels were linked to an increased risk of AKI. In addition, the established risk factor model for AKI can be considerably improved by the addition of serum magnesium in TAAR patients hospitalized in the CSICU.

Decoding the Mastectomy SKIN Score: An Evaluation of Its Predictive Performance in Immediate Breast Reconstruction

BACKGROUND

The skin ischemia and necrosis (SKIN) score was introduced to standardize the assessment of mastectomy skin flap necrosis (MSFN) severity and the need for reoperation. The authors evaluated the association between the SKIN score and the long-term postoperative outcomes of MSFN after mastectomy and immediate breast reconstruction.

METHODS

The authors conducted a retrospective cohort study of consecutive patients who developed MSFN after mastectomy and immediate breast reconstruction from January of 2001 to January of 2021. The primary outcome was breast-related complications after MSFN. Secondary outcomes were 30-day readmission, operating room (OR) débridement, and reoperation. Study outcomes were correlated with the SKIN composite score.

RESULTS

The authors identified 299 reconstructions in 273 consecutive patients with mean follow-up time of 111.8 ± 3.9 months. Most patients had a composite SKIN score of B2 (25.0%, n = 13), followed by D2 (17.3%) and C2 (15.4%). We found no significant difference in rates of OR débridement ( P = 0.347), 30-day readmission ( P = 0.167), any complication ( P = 0.492), or reoperation for a complication ( P = 0.189) based on the SKIN composite score. The composite skin score was a poor predictor of reoperation, with an area under the curve of 0.56. A subgroup analysis in patients who underwent implant-based reconstruction revealed no difference in rates of OR débridement ( P = 0.986), 30-day readmission ( P = 0.530), any complication ( P = 0.492), or reoperation for a complication ( P = 0.655) based on the SKIN composite score.

CONCLUSIONS

The SKIN score was a poor predictor for postoperative MSFN outcomes and reoperation. An individualized risk-assessment tool that incorporates the anatomic appearance of the breast, imaging data, and patient-level risk factors is needed.

CLINICAL QUESTION/LEVEL OF EVIDENCE

Risk, IV.

Urinary cytokeratin 20 as a predictor for chronic kidney disease following acute kidney injury

BACKGROUNDIdentifying patients with acute kidney injury (AKI) at high risk of chronic kidney disease (CKD) progression remains a challenge.METHODSKidney transcriptome sequencing was applied to identify the top upregulated genes in mice with AKI. The product of the top-ranking gene was identified in tubular cells and urine in mouse and human AKI. Two cohorts of patients with prehospitalization estimated glomerular filtration rate (eGFR) ≥ 45 mL/min/1.73 m2 who survived over 90 days after AKI were used to derive and validate the predictive models. AKI-CKD progression was defined as eGFR < 60 mL/min/1.73 m2 and with minimum 25% reduction from baseline 90 days after AKI in patients with prehospitalization eGFR ≥ 60 mL/min/1.73 m2. AKI-advanced CKD was defined as eGFR < 30 mL/min/1.73 m2 90 days after AKI in those with prehospitalization eGFR 45-59 mL/min/1.73 m2.RESULTSKidney cytokeratin 20 (CK20) was upregulated in injured proximal tubular cells and detectable in urine within 7 days after AKI. High concentrations of urinary CK20 (uCK20) were independently associated with the severity of histological AKI and the risk of AKI-CKD progression. In the Test set, the AUC of uCK20 for predicting AKI-CKD was 0.80, outperforming reported biomarkers for predicting AKI. Adding uCK20 to clinical variables improved the ability to predict AKI-CKD progression, with an AUC of 0.90, and improved the risk reclassification.CONCLUSIONThese findings highlight uCK20 as a useful predictor for AKI-CKD progression and may provide a tool to identify patients at high risk of CKD following AKI.FUNDINGNational Natural Science Foundation of China, National Key R&D Program of China, 111 Plan, Guangdong Key R&D Program.

Predictive Models for Assessing Patients' Response to Treatment in Metastatic Prostate Cancer: A Systematic Review

Background and objective

The treatment landscape of metastatic prostate cancer (mPCa) has evolved significantly over the past two decades. Despite this, the optimal therapy for patients with mPCa has not been determined. This systematic review identifies available predictive models that assess mPCa patients' response to treatment.

Methods

We critically reviewed MEDLINE and CENTRAL in December 2022 according to the Preferred Reporting Items for Systematic Reviews and Meta-analyses statement. Only quantitative studies in English were included with no time restrictions. The quality of the included studies was assessed using the PROBAST tool. Data were extracted following the Checklist for Critical Appraisal and Data Extraction for Systematic Reviews criteria.

Key findings and limitations

The search identified 616 citations, of which 15 studies were included in our review. Nine of the included studies were validated internally or externally. Only one study had a low risk of bias and a low risk concerning applicability. Many studies failed to detail model performance adequately, resulting in a high risk of bias. Where reported, the models indicated good or excellent performance.

Conclusions and clinical implications

Most of the identified predictive models require additional evaluation and validation in properly designed studies before these can be implemented in clinical practice to assist with treatment decision-making for men with mPCa.

Patient summary

In this review, we evaluate studies that predict which treatments will work best for which metastatic prostate cancer patients. We found that existing studies need further improvement before these can be used by health care professionals.

Development and Validation of an Intracranial Hemorrhage Risk Score in Older Adults with Atrial Fibrillation Treated with Oral Anticoagulant

Background

High risk of intracranial hemorrhage (ICH) is a leading reason for withholding anticoagulation in patients with atrial fibrillation (AF). We aimed to develop a claims-based ICH risk prediction model in older adults with AF initiating oral anticoagulation (OAC).

Methods

We used US Medicare claims data to identify new users of OAC aged ≥65 years with AF in 2010-2017. We used regularized Cox regression to select predictors of ICH. We compared our AF ICH risk score with the HAS-BLED bleed risk and Homer fall risk scores by area under the receiver operating characteristic curve (AUC) and assessed net reclassification improvement (NRI) when predicting 1-year risk of ICH.

Results

Our study cohort comprised 840,020 patients (mean [SD] age 77.5 [7.4] years and female 52.2%) split geographically into training (3963 ICH events [0.6%] in 629,804 patients) and validation (1397 ICH events [0.7%] in 210,216 patients) sets. Our AF ICH risk score, including 50 predictors, had superior AUCs of 0.653 and 0.650 in the training and validation sets than the HAS-BLED score of 0.580 and 0.567 (p<0.001) and the Homer score of 0.624 and 0.623 (p<0.001). In the validation set, our AF ICH risk score reclassified 57.8%, 42.5%, and 43.9% of low, intermediate, and high-risk patients, respectively, by HAS-BLED score (NRI: 15.3%, p<0.001). Similarly, it reclassified 0.0, 44.1, and 19.4% of low, intermediate, and high-risk patients, respectively, by the Homer score (NRI: 21.9%, p<0.001).

Conclusion

Our novel claims-based ICH risk prediction model outperformed the standard HAS-BLED score and can inform OAC prescribing decisions.

Hybridizing mechanistic mathematical modeling with deep learning methods to predict individual cancer patient survival after immune checkpoint inhibitor therapy

We present a study where predictive mechanistic modeling is used in combination with deep learning methods to predict individual patient survival probabilities under immune checkpoint inhibitor (ICI) therapy. This hybrid approach enables prediction based on both measures that are calculable from mechanistic models (but may not be directly measurable in the clinic) and easily measurable quantities or characteristics (that are not always readily incorporated into predictive mechanistic models). The mechanistic model we have applied here can predict tumor response from CT or MRI imaging based on key mechanisms underlying checkpoint inhibitor therapy, and in the present work, its parameters were combined with readily-available clinical measures from 93 patients into a hybrid training set for a deep learning time-to-event predictive model. Analysis revealed that training an artificial neural network with both mechanistic modeling-derived and clinical measures achieved higher per-patient predictive accuracy based on event-time concordance, Brier score, and negative binomial log-likelihood-based criteria than when only mechanistic model-derived values or only clinical data were used. Feature importance analysis revealed that both clinical and model-derived parameters play prominent roles in neural network decision making, and in increasing prediction accuracy, further supporting the advantage of our hybrid approach. We anticipate that many existing mechanistic models may be hybridized with deep learning methods in a similar manner to improve predictive accuracy through addition of additional data that may not be readily implemented in mechanistic descriptions.

Endothelium-related biomarkers enhanced prediction of kidney support therapy in critically ill patients with non-oliguric acute kidney injury

Acute kidney injury (AKI) is a common condition in hospitalized patients who often requires kidney support therapy (KST). However, predicting the need for KST in critically ill patients remains challenging. This study aimed to analyze endothelium-related biomarkers as predictors of KST need in critically ill patients with stage 2 AKI. A prospective observational study was conducted on 127 adult ICU patients with stage 2 AKI by serum creatinine only. Endothelium-related biomarkers, including vascular cell adhesion protein-1 (VCAM-1), angiopoietin (AGPT) 1 and 2, and syndecan-1, were measured. Clinical parameters and outcomes were recorded. Logistic regression models, receiver operating characteristic (ROC) curves, continuous net reclassification improvement (NRI) and integrated discrimination improvement (IDI) were used for analysis. Among the patients, 22 (17.2%) required KST within 72 h. AGPT2 and syndecan-1 levels were significantly greater in patients who progressed to the KST. Multivariate analysis revealed that AGPT2 and syndecan-1 were independently associated with the need for KST. The area under the ROC curve (AUC-ROC) for AGPT2 and syndecan-1 performed better than did the constructed clinical model in predicting KST. The combination of AGPT2 and syndecan-1 improved the discrimination capacity of predicting KST beyond that of the clinical model alone. Additionally, this combination improved the classification accuracy of the NRI and IDI. AGPT2 and syndecan-1 demonstrated predictive value for the need for KST in critically ill patients with stage 2 AKI. The combination of AGPT2 and syndecan-1 alone enhanced the predictive capacity of predicting KST beyond clinical variables alone. These findings may contribute to the early identification of patients who will benefit from KST and aid in the management of AKI in critically ill patients.

Development and Internal Validation of a Model for Predicting Overall Survival in Subjects with MAFLD: A Cohort Study

Background & Aims: Fatty liver disease with metabolic dysfunction (MAFLD) is a new concept proposed to replace the previous concept of Non-Alcoholic Hepatic Steatosis (NAFLD). We developed and internally validated a prognostic model to predict the likelihood of death in a cohort of subjects with MAFLD. Methods: Our work involved two steps: the first was the construction of a bootstrapped multivariable Cox model for mortality risk prognosis and the second was its validation. Results: The study cohort included 1506 subjects, of which 907 were used for internal validation. Discriminant measures for the final model were R2D 0.6845 and Harrell's C 0.8422 in the development and R2D 0.6930 and Harrell's C 0.8465 in the validation. We used the nine independent prognostic factors selected by the LASSO Cox procedure and fitted by the bootstrap Cox survival model, and observed β were: Gender 0.356 1.42 (p < 0.008), Age 0.146 (p < 0.001), Glycemia 0.004 (p < 0.002), Total Cholesterol -0.0040 (p < 0.009), Gamma Glutamyl Transpeptidase 0.009 (p < 0.001), SBP 0.009 (p < 0.036), DBP -0.016 (p < 0.041), ALP 0.008 (p < 0.071) and Widowhood 0.550 (p < 0.001). Conclusions: We produced and validated a model to estimate the probability of death in subjects with MAFLD. The instruments we used showed satisfactory predictive capabilities.

Clinical Predictive Modeling of Heart Failure: Domain Description, Models' Characteristics and Literature Review

This study attempts to identify and briefly describe the current directions in applied and theoretical clinical prediction research. Context-rich chronic heart failure syndrome (CHFS) telemedicine provides the medical foundation for this effort. In the chronic stage of heart failure, there are sudden exacerbations of syndromes with subsequent hospitalizations, which are called acute decompensation of heart failure (ADHF). These decompensations are the subject of diagnostic and prognostic predictions. The primary purpose of ADHF predictions is to clarify the current and future health status of patients and subsequently optimize therapeutic responses. We proposed a simplified discrete-state disease model as an attempt at a typical summarization of a medical subject before starting predictive modeling. The study tries also to structure the essential common characteristics of quantitative models in order to understand the issue in an application context. The last part provides an overview of prediction works in the field of CHFS. These three parts provide the reader with a comprehensive view of quantitative clinical predictive modeling in heart failure telemedicine with an emphasis on several key general aspects. The target community is medical researchers seeking to align their clinical studies with prognostic or diagnostic predictive modeling, as well as other predictive researchers. The study was written by a non-medical expert.

A study protocol for a predictive model to assess population-based avoidable hospitalization risk: Avoidable Hospitalization Population Risk Prediction Tool (AvHPoRT)

INTRODUCTION

Avoidable hospitalizations are considered preventable given effective and timely primary care management and are an important indicator of health system performance. The ability to predict avoidable hospitalizations at the population level represents a significant advantage for health system decision-makers that could facilitate proactive intervention for ambulatory care-sensitive conditions (ACSCs). The aim of this study is to develop and validate the Avoidable Hospitalization Population Risk Tool (AvHPoRT) that will predict the 5-year risk of first avoidable hospitalization for seven ACSCs using self-reported, routinely collected population health survey data.

METHODS AND ANALYSIS

The derivation cohort will consist of respondents to the first 3 cycles (2000/01, 2003/04, 2005/06) of the Canadian Community Health Survey (CCHS) who are 18-74 years of age at survey administration and a hold-out data set will be used for external validation. Outcome information on avoidable hospitalizations for 5 years following the CCHS interview will be assessed through data linkage to the Discharge Abstract Database (1999/2000-2017/2018) for an estimated sample size of 394,600. Candidate predictor variables will include demographic characteristics, socioeconomic status, self-perceived health measures, health behaviors, chronic conditions, and area-based measures. Sex-specific algorithms will be developed using Weibull accelerated failure time survival models. The model will be validated both using split set cross-validation and external temporal validation split using cycles 2000-2006 compared to 2007-2012. We will assess measures of overall predictive performance (Nagelkerke R2), calibration (calibration plots), and discrimination (Harrell's concordance statistic). Development of the model will be informed by the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD) statement.

ETHICS AND DISSEMINATION

This study was approved by the University of Toronto Research Ethics Board. The predictive algorithm and findings from this work will be disseminated at scientific meetings and in peer-reviewed publications.

Predicting Risk of 30-day Postoperative Morbidity Using the Pathologic Fracture Mortality Index

INTRODUCTION

The purpose of this study was to evaluate the ability of the Pathologic Fracture Mortality Index (PFMI) to predict the risk of 30-day morbidity after pathologic fracture fixation and compare its efficacy with those of the American Society of Anesthesiologists (ASA) physical status, modified Charlson Comorbidity Index (mCCI), and modified frailty index (mFI-5).

METHODS

Cohorts of 1,723 patients in the American College of Surgeons National Surgical Quality Improvement Program database from 2005 to 2020 and 159 patients from a tertiary cancer referral center who underwent fixation for impending or completed pathologic fractures of long bones were retrospectively analyzed. National Surgical Quality Improvement Program morbidity variables were categorized into medical, surgical, utilization, and all-cause. PFMI, ASA, mCCI, and mFI-5 scores were calculated for each patient. Area under the curve (AUC) was used to compare efficacies.

RESULTS

AUCs predicting all-cause morbidity were 0.62, 0.54, and 0.56 for the PFMI, ASA, and mFI-5, respectively. The PFMI outperformed the ASA and mFI-5 in predicting all-cause ( P < 0.01), medical ( P = 0.01), and utilization ( P < 0.01) morbidities. In the 2005 to 2012 subset, the PFMI outperformed the ASA, mFI-5, and mCCI in predicting all-cause ( P = 0.01), medical ( P = 0.03), and surgical ( P = 0.05) morbidities but performed similarly to utilization morbidity ( P = 0.19). In our institutional cohort, the AUC for the PFMI in morbidity stratification was 0.68. The PFMI was associated with all-cause (odds ratio [OR], 1.30; 95% confidence interval [CI], 1.12 to 1.51; P < 0.001), medical (OR, 1.19; 95% CI, 1.03 to 1.40; P = 0.046), and utilization (OR, 1.32; 95% CI, 1.14 to 1.52; P < 0.001) morbidities but not significantly associated with surgical morbidity (OR, 1.21; 95% CI, 0.98 to 1.49; P = 0.08) in this cohort.

DISCUSSION

The PFMI is an advancement in postoperative morbidity risk stratification of patients with pathologic fracture from metastatic disease.

LEVEL OF EVIDENCE

III.

Renal Biomarkers in Heart Failure: Systematic Review and Meta-Analysis

Background

Cystatin C, neutrophil gelatinase-associated lipocalin (NGAL), and kidney injury molecule (KIM)-1 are renal biomarkers increasingly appreciated for their role in the risk stratification and prognostication of heart failure (HF) patients. However, very few have been adopted clinically, owing to the lack of consistency.

Objectives

The authors aimed to study the association between cystatin C, NGAL, and KIM-1 and outcomes, mortality, hospitalizations, and worsening renal function (WRF) in patients with acute and chronic HF.

Methods

We included peer-reviewed English-language articles from PubMed and EMBASE published up to December 2021. We analyzed the above associations using random-effects meta-analysis. Publication bias was assessed using funnel plots.

Results

Among 2,631 articles, 100 articles, including 45,428 patients, met the inclusion criteria. Top-tertile of serum cystatin C, when compared to the bottom-tertile, carried a higher pooled hazard ratio (pHR) for mortality (pHR: 1.59, 95% CI: 1.42-1.77) and for the composite outcome of mortality and HF hospitalizations (pHR: 1.49, 95% CI: 1.23-1.75). Top-tertile of serum NGAL had a higher hazard for mortality (pHR: 2.91, 95% CI: 1.49-5.67) and composite outcome (HR: 4.11, 95% CI: 2.69-6.30). Serum and urine NGAL were significantly associated with WRF, with pHRs of 2.40 (95% CI: 1.48-3.90) and 2.01 (95% CI: 1.21-3.35). Urine KIM-1 was significantly associated with WRF (pHR: 1.60, 95% CI: 1.24-2.07) but not with other outcomes. High heterogeneity was noted between studies without an obvious explanation based on meta-regression.

Conclusions

Serum cystatin C and serum NGAL are independent predictors of adverse outcomes in HF. Serum and urine NGAL are important predictors of WRF in HF.

Towards the Clinical Implementation of Extracellular Vesicle-Based Biomarker Assays for Cancer

BACKGROUND

Substantial research has been devoted to elucidating the role of extracellular vesicles (EVs) in the different hallmarks of cancer. Consequently, EVs are increasingly explored as a source of cancer biomarkers in body fluids. However, the heterogeneity in EVs, the complexity of body fluids, and the diversity in methods available for EV analysis, challenge the development and translation of EV-based biomarker assays.

CONTENT

Essential steps in EV-associated biomarker development are emphasized covering biobanking, biomarker discovery, verification and validation, and clinical implementation. A meticulous study design is essential and ideally results from close interactions between clinicians and EV researchers. A plethora of different EV preparation protocols exists which warrants quality control and transparency to ensure reproducibility and thus enable verification of EV-associated biomarker candidates identified in the discovery phase in subsequent independent cohorts. The development of an EV-associated biomarker assay requires thorough analytical and clinical validation. Finally, regulatory affairs must be considered for clinical implementation of EV-based biomarker assays.

SUMMARY

In this review, the current challenges that prevent us from exploiting the full potential of EV-based biomarker assays are identified. Guidelines and tools to overcome these hurdles are highlighted and are crucial to advance EV-based biomarker assays into clinical use.

External validation of models for predicting cumulative live birth over multiple complete cycles of IVF treatment

STUDY QUESTION

Can two prediction models developed using data from 1999 to 2009 accurately predict the cumulative probability of live birth per woman over multiple complete cycles of IVF in an updated UK cohort?

SUMMARY ANSWER

After being updated, the models were able to estimate individualized chances of cumulative live birth over multiple complete cycles of IVF with greater accuracy.

WHAT IS KNOWN ALREADY

The McLernon models were the first to predict cumulative live birth over multiple complete cycles of IVF. They were converted into an online calculator called OPIS (Outcome Prediction In Subfertility) which has 3000 users per month on average. A previous study externally validated the McLernon models using a Dutch prospective cohort containing data from 2011 to 2014. With changes in IVF practice over time, it is important that the McLernon models are externally validated on a more recent cohort of patients to ensure that predictions remain accurate.

STUDY DESIGN, SIZE, DURATION

A population-based cohort of 91 035 women undergoing IVF in the UK between January 2010 and December 2016 was used for external validation. Data on frozen embryo transfers associated with these complete IVF cycles conducted from 1 January 2017 to 31 December 2017 were also collected.

PARTICIPANTS/MATERIALS, SETTING, METHODS

Data on IVF treatments were obtained from the Human Fertilisation and Embryology Authority (HFEA). The predictive performances of the McLernon models were evaluated in terms of discrimination and calibration. Discrimination was assessed using the c-statistic and calibration was assessed using calibration-in-the-large, calibration slope, and calibration plots. Where any model demonstrated poor calibration in the validation cohort, the models were updated using intercept recalibration, logistic recalibration, or model revision to improve model performance.

MAIN RESULTS AND THE ROLE OF CHANCE

Following exclusions, 91 035 women who underwent 144 734 complete cycles were included. The validation cohort had a similar distribution age profile to women in the development cohort. Live birth rates over all complete cycles of IVF per woman were higher in the validation cohort. After calibration assessment, both models required updating. The coefficients of the pre-treatment model were revised, and the updated model showed reasonable discrimination (c-statistic: 0.67, 95% CI: 0.66 to 0.68). After logistic recalibration, the post-treatment model showed good discrimination (c-statistic: 0.75, 95% CI: 0.74 to 0.76). As an example, in the updated pre-treatment model, a 32-year-old woman with 2 years of primary infertility has a 42% chance of having a live birth in the first complete ICSI cycle and a 77% chance over three complete cycles. In a couple with 2 years of primary male factor infertility where a 30-year-old woman has 15 oocytes collected in the first cycle, a single fresh blastocyst embryo transferred in the first cycle and spare embryos cryopreserved, the estimated chance of live birth provided by the post-treatment model is 46% in the first complete ICSI cycle and 81% over three complete cycles.

LIMITATIONS, REASONS FOR CAUTION

Two predictors from the original models, duration of infertility and previous pregnancy, which were not available in the recent HFEA dataset, were imputed using data from the older cohort used to develop the models. The HFEA dataset does not contain some other potentially important predictors, e.g. BMI, ethnicity, race, smoking and alcohol intake in women, as well as measures of ovarian reserve such as antral follicle count.

WIDER IMPLICATIONS OF THE FINDINGS

Both updated models show improved predictive ability and provide estimates which are more reflective of current practice and patient case mix. The updated OPIS tool can be used by clinicians to help shape couples' expectations by informing them of their individualized chances of live birth over a sequence of multiple complete cycles of IVF.

STUDY FUNDING/COMPETING INTEREST(S)

This study was supported by an Elphinstone scholarship scheme at the University of Aberdeen and Aberdeen Fertility Centre, University of Aberdeen. S.B. has a commitment of research funding from Merck. D.J.M. and M.B.R. declare support for the present manuscript from Elphinstone scholarship scheme at the University of Aberdeen and Assisted Reproduction Unit at Aberdeen Fertility Centre, University of Aberdeen. D.J.M. declares grants received by University of Aberdeen from NHS Grampian, The Meikle Foundation, and Chief Scientist Office in the past 3 years. D.J.M. declares receiving an honorarium for lectures from Merck. D.J.M. is Associate Editor of Human Reproduction Open and Statistical Advisor for Reproductive BioMed Online. S.B. declares royalties from Cambridge University Press for a book. S.B. declares receiving an honorarium for lectures from Merck, Organon, Ferring, Obstetric and Gynaecological Society of Singapore, and Taiwanese Society for Reproductive Medicine. S.B. has received support from Merck, ESHRE, and Ferring for attending meetings as speaker and is on the METAFOR and CAPRE Trials Data Monitoring Committee.

TRIAL REGISTRATION NUMBER

N/A.

Tumor budding in breast carcinoma: A systematic review and meta-analysis

ABSTRACT

Tumor budding is gaining importance as a prognostic factor in various carcinomas due to its association with epithelial-mesenchymal transition (EMT) and hence clinical outcome. Reporting tumor budding in breast cancer lacks homogeneity. We aim to systematically review the existing literature and conduct a meta-analysis to assess the prognostic implication of tumor budding in breast carcinoma. A systematic search was performed to identify studies that compared different prognostic variables between high- and low-grade tumor budding. Quality assessment was performed using a modified Newcastle Ottawa Scale. Dichotomous variables were pooled using the odds ratio using the Der-Simonian-Laird method. Meta-analysis was conducted to study the association between low/high-grade tumor budding and tumor grade, lymph node metastasis, lymphovascular invasion, ER, PR, HER2neu, KI67, and the molecular subtype triple-negative breast carcinoma. Thirteen studies with a total of 1763 patients were included. A moderate risk of bias was noted. The median bias scoring was 7 (6-9). High-grade tumor budding was significantly associated with lymph node metastasis (OR: 2.25, 95% CI: 1.52-3.34, P < 0.01) and lymphovascular invasion (OR: 3.14, 95% CI: 2.10-4.71, P < 0.01), and low-grade budding was significantly associated with triple-negative breast carcinoma (OR: 0.61, 95% CI: 0.39-0.95, P = 0.03)There was significant heterogeneity in the assessment and grading of tumor budding; thus, a checklist of items was identified that lacked standardization. Our meta-analysis concluded that tumor budding can act as an independent prognostic marker for breast cancer.

A generalizable and interpretable model for mortality risk stratification of sepsis patients in intensive care unit

PURPOSE

This study aimed to construct a mortality model for the risk stratification of intensive care unit (ICU) patients with sepsis by applying a machine learning algorithm.

METHODS

Adult patients who were diagnosed with sepsis during admission to ICU were extracted from MIMIC-III, MIMIC-IV, eICU, and Zigong databases. MIMIC-III was used for model development and internal validation. The other three databases were used for external validation. Our proposed model was developed based on the Extreme Gradient Boosting (XGBoost) algorithm. The generalizability, discrimination, and validation of our model were evaluated. The Shapley Additive Explanation values were used to interpret our model and analyze the contribution of individual features.

RESULTS

A total of 16,741, 15,532, 22,617, and 1,198 sepsis patients were extracted from the MIMIC-III, MIMIC-IV, eICU, and Zigong databases, respectively. The proposed model had an area under the receiver operating characteristic curve (AUROC) of 0.84 in the internal validation, which outperformed all the traditional scoring systems. In the external validations, the AUROC was 0.87 in the MIMIC-IV database, better than all the traditional scoring systems; the AUROC was 0.83 in the eICU database, higher than the Simplified Acute Physiology Score III and Sequential Organ Failure Assessment (SOFA),equal to 0.83 of the Acute Physiology and Chronic Health Evaluation IV (APACHE-IV), and the AUROC was 0.68 in the Zigong database, higher than those from the systemic inflammatory response syndrome and SOFA. Furthermore, the proposed model showed the best discriminatory and calibrated capabilities and had the best net benefit in each validation.

CONCLUSIONS

The proposed algorithm based on XGBoost and SHAP-value feature selection had high performance in predicting the mortality of sepsis patients within 24 h of ICU admission.

How the Oviduct Lipidomic Profile Changes over Time after the Start of an Obesogenic Diet in an Outbred Mouse Model

We investigated whether a high-fat/high-sugar (HF/HS) diet alters the lipidomic profile of the oviductal epithelium (OE) and studied the patterns of these changes over time. Female outbred Swiss mice were fed either a control (10% fat) or HF/HS (60% fat, 20% fructose) diet. Mice (n = 3 per treatment per time point) were sacrificed and oviducts were collected at 3 days and 1, 4, 8, 12 and 16 weeks on the diet. Lipids in the OE were imaged using matrix-assisted laser desorption ionisation mass spectrometry imaging. Discriminative m/z values and differentially regulated lipids were determined in the HF/HS versus control OEs at each time point. Feeding the obesogenic diet resulted in acute changes in the lipid profile in the OE already after 3 days, and thus even before the development of an obese phenotype. The changes in the lipid profile of the OE progressively increased and became more persistent after long-term HF/HS diet feeding. Functional annotation revealed a differential abundance of phospholipids, sphingomyelins and lysophospholipids in particular. These alterations appear to be not only caused by the direct accumulation of the excess circulating dietary fat but also a reduction in the de novo synthesis of several lipid classes, due to oxidative stress and endoplasmic reticulum dysfunction. The described diet-induced lipidomic changes suggest alterations in the OE functions and the oviductal microenvironment which may impact crucial reproductive events that take place in the oviduct, such as fertilization and early embryo development.

Syndecan-1 predicts hemodynamic instability in critically ill patients under intermittent hemodialysis

Introduction

Up to 70% of intermittent hemodialysis (IHD) sessions in critically ill patients are complicated by hemodynamic instability. Although several clinical characteristics have been associated with hemodynamic instability during IHD, the discriminatory capacity of predicting such events during IHD sessions is less defined. In the present study, we aimed to analyse endothelium-related biomarkers collected before IHD sessions and their capacity to predict hemodynamic instability related to IHD in critically ill patients.

Methods

In this prospective observational study, we enrolled adult critically ill patients with acute kidney injury who required fluid removal with IHD. We screened each included patient daily for IHD sessions. Thirty minutes before each IHD session, each patient had a 5-mL blood collection for measurement of endothelial biomarkers-vascular cell adhesion molecule-1 (VCAM-1), angiopoietin-1 and -2 (AGPT1 and AGPT2) and syndecan-1. Hemodynamic instability during IHD was the main outcome. Analyses were adjusted for variables already known to be associated with hemodynamic instability during IHD.

Results

Plasma syndecan-1 was the only endothelium-related biomarker independently associated with hemodynamic instability. The accuracy of syndecan-1 for predicting hemodynamic instability during IHD was moderate [area under the receiver operating characteristic curve 0.78 (95% confidence interval 0.68-0.89)]. The addition of syndecan-1 improved the discrimination capacity of a clinical model from 0.67 to 0.82 (P < .001) and improved risk prediction, as measured by net reclassification improvement.

Conclusion

Syndecan-1 is associated with hemodynamic instability during IHD in critically ill patients. It may be useful to identify patients who are at increased risk for such events and suggests that endothelial glycocalyx derangement is involved in the pathophysiology of IHD-related hemodynamic instability.

The Association Between Enhanced Recovery After Cardiac Surgery-Guided Analgesics and Postoperative Delirium

OBJECTIVES

Delirium is a common postoperative complication associated with death and long-term cognitive impairment. The authors studied the association between opioid-sparing anesthetics, incorporating Enhanced Recovery After Cardiac Surgery (ERACS)-guided analgesics and postoperative delirium.

DESIGN

The authors performed a retrospective review of nonemergent coronary, valve, or ascending aorta surgery patients.

SETTING

A tertiary academic medical institution.

PARTICIPANTS

The study authors analyzed a dataset of elective adult cardiac surgical patients. All patients ≥18 years undergoing elective cardiac surgery from November 2, 2017 until February 2, 2021 were eligible for inclusion.

INTERVENTIONS

The ERACS-guided multimodal pain regimen included preoperative oral acetaminophen and gabapentin, and intraoperative intravenous lidocaine, ketamine, and dexmedetomidine.

MEASUREMENTS AND MAIN RESULTS

Delirium was measured by bedside nurses using the Confusion Assessment Method for the intensive care unit (ICU). Delirium occurred in 220 of the 1,675 patients (13.7%). The use of any component of the multimodal pain regimen was not associated with delirium (odds ratio [OR]: 0.85 [95% CI: 0.63-1.16]). Individually, acetaminophen was associated with reduced odds of delirium (OR: 0.60 [95% CI: 0.37-0.95]). Gabapentin (OR: 1.36 [95% CI: 0.97-2.21]), lidocaine (OR: 0.86 [95% CI: 0.53-1.37]), ketamine (OR: 1.15 [95% CI: 0.72-1.83]), and dexmedetomidine (OR: 0.79 [95% CI: 0.46-1.31]) were not individually associated with postoperative delirium. Individual ERACS elements were associated with secondary outcomes of hospital length of stay, ICU duration, postoperative opioid administration, and postoperative intubation duration.

CONCLUSIONS

The use of an opioid-sparing perioperative ERACS pain regimen was not associated with reduced postoperative delirium, opioid consumption, or additional poor outcomes. Individually, acetaminophen was associated with reduced delirium.

LOCC: a novel visualization and scoring of cutoffs for continuous variables

Objective

There is a need for new methods to select and analyze cutoffs employed to define genes that are most prognostic significant and impactful. We designed LOCC (Luo's Optimization Categorization Curve), a novel tool to visualize and score continuous variables for a dichotomous outcome.

Methods

To demonstrate LOCC with real world data, we analyzed TCGA hepatocellular carcinoma gene expression and patient data using LOCC. We compared LOCC visualization to receiver operating characteristic (ROC) curve for prognostic modeling to showcase its utility in understanding predictors in various TCGA datasets.

Results

Analysis of E2F1 expression in hepatocellular carcinoma using LOCC demonstrated appropriate cutoff selection and validation. In addition, we compared LOCC visualization and scoring to ROC curves and c-statistics, demonstrating that LOCC better described predictors. Analysis of a previously published gene signature showed large differences in LOCC scoring, and removing the lowest scoring genes did not affect prognostic modeling of the gene signature demonstrating LOCC scoring could distinguish which predictors were most critical.

Conclusion

Overall, LOCC is a novel visualization tool for understanding and selecting cutoffs, particularly for gene expression analysis in cancer. The LOCC score can be used to rank genes for prognostic potential and is more suitable than ROC curves for prognostic modeling.

Evaluation of a Structured Screening Assessment to Detect Isolated Rapid Eye Movement Sleep Behavior Disorder

BACKGROUND

Isolated rapid eye movement (REM) sleep behavior disorder (iRBD) cohorts have provided insights into the earliest neurodegenerative processes in α-synucleinopathies. Even though polysomnography (PSG) remains the gold standard for diagnosis, an accurate questionnaire-based algorithm to identify eligible subjects could facilitate efficient recruitment in research.

OBJECTIVE

This study aimed to optimize the identification of subjects with iRBD from the general population.

METHODS

Between June 2020 and July 2021, we placed newspaper advertisements, including the single-question screen for RBD (RBD1Q). Participants' evaluations included a structured telephone screening consisting of the RBD screening questionnaire (RBDSQ) and additional sleep-related questionnaires. We examined anamnestic information predicting PSG-proven iRBD using logistic regressions and receiver operating characteristic curves.

RESULTS

Five hundred forty-three participants answered the advertisements, and 185 subjects fulfilling inclusion and exclusion criteria were screened. Of these, 124 received PSG after expert selection, and 78 (62.9%) were diagnosed with iRBD. Selected items of the RBDSQ, the Pittsburgh Sleep Quality Index, the STOP-Bang questionnaire, and age predicted iRBD with high accuracy in a multiple logistic regression model (area under the curve >80%). When comparing the algorithm to the sleep expert decision, 77 instead of 124 polysomnographies (62.1%) would have been carried out, and 63 (80.8%) iRBD patients would have been identified; 32 of 46 (69.6%) unnecessary PSG examinations could have been avoided.

CONCLUSIONS

Our proposed algorithm displayed high diagnostic accuracy for PSG-proven iRBD cost-effectively and may be a convenient tool for research and clinical settings. External validation sets are warranted to prove reliability. © 2023 The Authors. Movement Disorders published by Wiley Periodicals LLC on behalf of International Parkinson and Movement Disorder Society.

Relationship of Fibroblast Growth Factor 23 With Hospitalization for Heart Failure and Cardiovascular Outcomes in Patients Undergoing Cardiac Surgery

Background Fibroblast growth factor 23 (FGF-23) is crucial in regulating phosphate and vitamin D metabolism and is moreover associated with an increased cardiovascular risk. The specific objective of this study was to investigate the influence of FGF-23 on cardiovascular outcomes, including hospitalization for heart failure (HHF), postoperative atrial fibrillation, and cardiovascular death, in an unselected patient population after cardiac surgery. Methods and Results Patients undergoing elective coronary artery bypass graft and/or cardiac valve surgery were prospectively enrolled. FGF-23 blood plasma concentrations were assessed before surgery. A composite of cardiovascular death/HHF was chosen as primary end point. A total of 451 patients (median age 70 years; 28.8% female) were included in the present analysis and followed over a median of 3.9 years. Individuals with higher FGF-23 quartiles showed elevated incidence rates of the composite of cardiovascular death/HHF (quartile 1, 7.1%; quartile 2, 8.6%; quartile 3, 15.1%; and quartile 4, 34.3%). After multivariable adjustment, FGF-23 modeled as a continuous variable (adjusted hazard ratio for a 1-unit increase in standardized log-transformed biomarker, 1.82 [95% CI, 1.34-2.46]) as well as using predefined risk groups and quartiles remained independently associated with the risk of cardiovascular death/HHF and the secondary outcomes, including postoperative atrial fibrillation. Reclassification analysis indicated that the addition of FGF-23 to N-terminal pro-B-type natriuretic peptide provides a significant improvement in risk discrimination (net reclassification improvement at the event rate, 0.58 [95% CI, 0.34-0.81]; P<0.001; integrated discrimination increment, 0.03 [95% CI, 0.01-0.05]; P<0.001). Conclusions FGF-23 is an independent predictor of cardiovascular death/HHF and postoperative atrial fibrillation in individuals undergoing cardiac surgery. Considering an individualized risk assessment, routine preoperative FGF-23 evaluation may improve detection of high-risk patients.

Machine learning in the prediction of cardiac surgery associated acute kidney injury with early postoperative biomarkers

Purpose

To establish novel prediction models for predicting acute kidney injury (AKI) after cardiac surgery based on early postoperative biomarkers.

Patients and methods

This study enrolled patients who underwent cardiac surgery in a Chinese tertiary cardiac center and consisted of a discovery cohort (n = 452, from November 2018 to June 2019) and a validation cohort (n = 326, from December 2019 to May 2020). 43 biomarkers were screened using the least absolute shrinkage and selection operator and logistic regression to construct a nomogram model. Three tree-based machine learning models were also established: eXtreme Gradient Boosting (XGBoost), random forest (RF) and deep forest (DF). Model performance was accessed using area under the receiver operating characteristic curve (AUC). AKI was defined according to the Kidney Disease Improving Global Outcomes criteria.

Results

Five biomarkers were identified as independent predictors of AKI and were included in the nomogram: soluble ST2 (sST2), N terminal pro-brain natriuretic peptide (NT-proBNP), heart-type fatty acid binding protein (H-FABP), lactic dehydrogenase (LDH), and uric acid (UA). In the validation cohort, the nomogram achieved good discrimination, with AUC of 0.834. The machine learning models also exhibited adequate discrimination, with AUC of 0.856, 0.850, and 0.836 for DF, RF, and XGBoost, respectively. Both nomogram and machine learning models had well calibrated. The AUC of sST2, NT-proBNP, H-FABP, LDH, and UA to discriminate AKI were 0.670, 0.713, 0.725, 0.704, and 0.749, respectively. In addition, all of these biomarkers were significantly correlated with AKI after adjusting clinical confounders (odds ratio and 95% confidence interval of the third vs. the first tertile: sST2, 3.55 [2.34-5.49], NT-proBNP, 5.50 [3.54-8.71], H-FABP, 6.64 [4.11-11.06], LDH, 7.47 [4.54-12.64], and UA, 8.93 [5.46-15.06]).

Conclusion

Our study provides a series of novel predictive models and five biomarkers for enhancing the risk stratification of AKI after cardiac surgery.

Serum N-terminal pro-B-type natriuretic peptide and cystatin C for acute kidney injury detection in critically ill adults in China: a prospective, observational study

OBJECTIVE

Serum N-terminal pro-B-type natriuretic peptide (NT-proBNP) and cystatin C (sCysC) are available clinically and beneficial in diagnosing acute kidney injury (AKI). Our purpose is to identify the performance of their combined diagnosis for AKI in critically ill patients.

DESIGN

A prospectively recruited, observational study was performed.

SETTING

Adults admitted to the intensive care unit of a tertiary hospital in China.

PARTICIPANTS

A total of 1222 critically ill patients were enrolled in the study.

MAIN OUTCOME MEASURES

To identify the performance of the combined diagnosis of serum NT-proBNP and sCysC for AKI in critically ill patients. The area under the receiver operating characteristic curve (AUC-ROC), category-free net reclassification index (NRI) and incremental discrimination improvement (IDI) were utilised for comparing the discriminative powers of a combined and single biomarker adjusted model of clinical variables enriched with NT-proBNP and sCysC for AKI.

RESULTS

AKI was detected in 256 out of 1222 included patients (20.9%). AUC-ROC for NT-proBNP and sCysC to detect AKI had a significantly higher accuracy than any individual biomarker (p<0.05). After multivariate adjustment, a level of serum NT-proBNP ≥204 pg/mL was associated with 3.5-fold higher odds for AKI compared with those below the cut-off value. Similar results were obtained for sCysC levels (p<0.001). To detect AKI, adding NT-proBNP and sCysC to a clinical model further increased the AUC-ROC to 0.859 beyond that of the clinical model with or without sCysC (p<0.05). Moreover, the addition of these two to the clinical model significantly improved risk reclassification of AKI beyond that of the clinical model alone or with single biomarker (p<0.05), as measured by NRI and IDI.

CONCLUSIONS

In critically ill individuals, serum NT-proBNP, sCysC and clinical risk factors combination improve the discriminative power for diagnosing AKI.

Investigating the role of FADS family members in breast cancer based on bioinformatic analysis and experimental validation

Breast cancer (BC) is the most common malignant tumor in women worldwide. Emerging evidence indicates the significance of fatty acid metabolism in BC. Fatty acid desaturase (FADS) is closely associated with cancer occurrence and development. Here, bioinformatic analysis and experimental validation were applied to investigate the potential functions of FADS in BC. Several public databases, including TCGA, GEO, HPA, Kaplan-Meier plotter, STRING, DAVID, cBioPortal, TIMER, TRRUST, and LinkedOmics were used to determine mRNA/protein expression levels, prognostic significance, functional enrichment, genetic alterations, association with tumor-infiltrating immune cells, and related transcription factors and kinases. BC tissues showed higher and lower mRNA expression of FADS2/6/8 and FADS3/4/5, respectively. FADS1/2/6 and FADS3/4/5 showed higher and lower protein expression levels, respectively, in BC tissues. Moreover, FADS1/7 up- and FADS3/8 down-regulation predicted poor overall and recurrence-free survival, while FADS2/5 up- and FADS4 down-regulation were associated with poor recurrence-free survival. Receiver operating characteristic curves revealed that FADS2/3/4/8 were indicative diagnostic markers. FADS family members showing differential expression levels were associated with various clinical subtypes, clinical stages, lymph node metastasis status, copy number variants, DNA methylation, and miRNA regulation in BC. The mRNA expression level of FADS1/2/3/4/5/7/8 was observed to be significantly negatively correlated with DNA methylation. FADS1/2 upregulation was significantly correlated with clinical stages. FADS1/4 expression was obviously lower in BC patients with higher lymph node metastasis than lower lymph node metastasis, while FADS7/8 expression was obviously higher in BC patients with higher lymph node metastasis than lower lymph node metastasis. FADS family members showed varying degrees of genetic alterations, and Gene Ontology and KEGG pathway enrichment analyses suggested their involvement in lipid metabolism. Their expression level was correlated with immune cell infiltration levels. FADS2 was chosen for further validation analyses. We found FADS2 to be significantly over-expressed in clinical BC tissue samples. The proliferation, migration, and invasion abilities of MDA-MB-231 and BT474 cells were significantly reduced after FADS2 knockdown. Furthermore, FADS2 may promote the occurrence and development of BC cells via regulating the epithelial-mesenchymal transition (EMT) pathway. Altogether, our results suggest that FADS1/2/3/4 can serve as potential therapeutic targets, prognostic indicators, and diagnostic markers in patients with BC.

Optimal cutoff scores of the Chinese version of 15-item negative symptom assessment that indicate prominent negative symptoms of schizophrenia

Objective

The Chinese version of 15-item negative symptom assessment (NSA-15) is an instrument with a three-factor structure specifically validated for assessing negative symptoms of schizophrenia. To provide a reference for future practical applications in the recognition of schizophrenia patients with negative symptoms, this study aimed to determine an appropriate NSA-15 cutoff score regarding negative symptoms to identify prominent negative symptoms (PNS).

Methods

A total of 199 participants with schizophrenia were recruited and divided into the PNS group (n = 79) and non-PNS group (n = 120) according to scale for assessment of negative symptoms (SANS) scores. Receiver-operating characteristic (ROC) curve analysis was used to determine the optimal NSA-15 cutoff score for identifying PNS.

Results

The optimal cutoff NSA-15 score for identifying PNS was 40. Communication, emotion and motivation factors in the NSA-15 had cutoffs of 13, 6, and 16, respectively. The communication factor score had slightly better discrimination than scores on the other two factors. The discriminant ability of the global rating of the NSA-15 was not as good as that of the NSA-15 total score (area under the curve (AUC): 0.873 vs. 0.944).

Conclusion

The optimal NSA-15 cutoff scores for identifying PNS in schizophrenia were determined in this study. The NSA-15 provides a convenient and easy-to-use assessment for identifying patients with PNS in Chinese clinical situations. The communication factor of the NSA-15 also has excellent discrimination.

A Surgeon's Guide to Artificial Intelligence-Driven Predictive Models

Artificial intelligence (AI) focuses on processing and interpreting complex information as well as identifying relationships and patterns among complex data. Artificial intelligence- and machine learning (ML)-driven predictions have shown promising potential in influencing real-time decisions and improving surgical outcomes by facilitating screening, diagnosis, risk assessment, preoperative planning, and shared decision-making. Fundamental understanding of the algorithms, as well as their development and interpretation, is essential for the evolution of AI in surgery. In this article, we provide surgeons with a fundamental understanding of AI-driven predictive models through an overview of common ML and deep learning algorithms, model development, performance metrics and interpretation. This would serve as a basis for understanding ML-based research, while fostering new ideas and innovations for furthering the reach of this emerging discipline.

Deep ROC Analysis and AUC as Balanced Average Accuracy, for Improved Classifier Selection, Audit and Explanation

Optimal performance is desired for decision-making in any field with binary classifiers and diagnostic tests, however common performance measures lack depth in information. The area under the receiver operating characteristic curve (AUC) and the area under the precision recall curve are too general because they evaluate all decision thresholds including unrealistic ones. Conversely, accuracy, sensitivity, specificity, positive predictive value and the F1 score are too specific-they are measured at a single threshold that is optimal for some instances, but not others, which is not equitable. In between both approaches, we propose deep ROC analysis to measure performance in multiple groups of predicted risk (like calibration), or groups of true positive rate or false positive rate. In each group, we measure the group AUC (properly), normalized group AUC, and averages of: sensitivity, specificity, positive and negative predictive value, and likelihood ratio positive and negative. The measurements can be compared between groups, to whole measures, to point measures and between models. We also provide a new interpretation of AUC in whole or part, as balanced average accuracy, relevant to individuals instead of pairs. We evaluate models in three case studies using our method and Python toolkit and confirm its utility.

Developing a prediction model for successful aging among the elderly using machine learning algorithms

Objective

The aging phenomenon has an increasing trend worldwide which caused the emergence of the successful aging (SA)1 concept. It is believed that the SA prediction model can increase the quality of life (QoL)2 in the elderly by decreasing physical and mental problems and enhancing their social participation. Most previous studies noted that physical and mental disorders affected the QoL in the elderly but didn't pay much attention to the social factors in this respect. Our study aimed to build a prediction model for SA based on the physical, mental, and specially more social factors affecting SA.

Methods

The 975 cases related to SA and non-SA of the elderly were investigated in this study. We used the univariate analysis to determine the best factors affecting the SA. AB3, XG-Boost J-48, RF4, artificial neural network5, support vector machine6, and NB7 algorithms were used for building the prediction models. To get the best model predicting the SA, we compared them using positive predictive value (PPV)8, negative predictive value (NPV)9, sensitivity, specificity, accuracy, F-measure, and area under the receiver operator characteristics curve (AUC).

Results

Comparing the machine learning10 model's performance showed that the random forest (RF) model with PPV = 90.96%, NPV = 99.21%, sensitivity = 97.48%, specificity = 97.14%, accuracy = 97.05%, F-score = 97.31%, AUC = 0.975 is the best model for predicting the SA.

Conclusions

Using prediction models can increase the QoL in the elderly and consequently reduce the economic cost for people and societies. The RF can be considered an optimal model for predicting SA in the elderly.

Calibrating violence risk assessments for uncertainty

Psychiatrists and other mental health clinicians are often tasked with assessing patients' risk of violence. Approaches to this vary and include both unstructured (based on individual clinicians' judgement) and structured methods (based on formalised scoring and algorithms with varying scope for clinicians' judgement). The end result is usually a categorisation of risk, which may, in turn, reference a probability estimate of violence over a certain time period. Research over recent decades has made considerable improvements in refining structured approaches and categorising patients' risk classifications at a group level. The ability, however, to apply these findings clinically to predict the outcomes of individual patients remains contested. In this article, we review methods of assessing violence risk and empirical findings on their predictive validity. We note, in particular, limitations in calibration (accuracy at predicting absolute risk) as distinct from discrimination (accuracy at separating patients by outcome). We also consider clinical applications of these findings, including challenges applying statistics to individual patients, and broader conceptual issues in distinguishing risk and uncertainty. Based on this, we argue that there remain significant limits to assessing violence risk for individuals and that this requires careful consideration in clinical and legal contexts.

Application of machine learning techniques for predicting survival in ovarian cancer

BACKGROUND

Ovarian cancer is the fifth leading cause of mortality among women in the United States. Ovarian cancer is also known as forgotten cancer or silent disease. The survival of ovarian cancer patients depends on several factors, including the treatment process and the prognosis.

METHODS

The ovarian cancer patients' dataset is compiled from the Surveillance, Epidemiology, and End Results (SEER) database. With the help of a clinician, the dataset is curated, and the most relevant features are selected. Pearson's second coefficient of skewness test is used to evaluate the skewness of the dataset. Pearson correlation coefficient is also used to investigate the associations between features. Statistical test is utilized to evaluate the significance of the features. Six Machine Learning (ML) models, including K-Nearest Neighbors , Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), Adaptive Boosting (AdaBoost), and Extreme Gradient Boosting (XGBoost), are implemented for survival prediction in both classification and regression approaches. An interpretable method, Shapley Additive Explanations (SHAP), is applied to clarify the decision-making process and determine the importance of each feature in prediction. Additionally, DTs of the RF model are displayed to show how the model predicts the survival intervals.

RESULTS

Our results show that RF (Accuracy = 88.72%, AUC = 82.38%) and XGBoost (Root Mean Squad Error (RMSE)) = 20.61%, R² = 0.4667) have the best performance for classification and regression approaches, respectively. Furthermore, using the SHAP method along with extracted DTs of the RF model, the most important features in the dataset are identified. Histologic type ICD-O-3, chemotherapy recode, year of diagnosis, age at diagnosis, tumor stage, and grade are the most important determinant factors in survival prediction.

CONCLUSION

To the best of our knowledge, our study is the first study that develops various ML models to predict ovarian cancer patients' survival on the SEER database in both classification and regression approaches. These ML algorithms also achieve more accurate results and outperform statistical methods. Furthermore, our study is the first study to use the SHAP method to increase confidence and transparency of the proposed models' prediction for clinicians. Moreover, our developed models, as an automated auxiliary tool, can help clinicians to have a better understanding of the estimated survival as well as important features that affect survival.

DNA methylation signatures as biomarkers of socioeconomic position

This review article provides a framework for the use of deoxyribonucleic acid (DNA) methylation (DNAm) biomarkers to study the biological embedding of socioeconomic position (SEP) and summarizes the latest developments in the area. It presents the emerging literature showing associations between individual- and neighborhood-level SEP exposures and DNAm across the life course. In contrast to questionnaire-based methods of assessing SEP, we suggest that DNAm biomarkers may offer an accessible metric to study questions about SEP and health outcomes, acting as a personal dosimeter of exposure. However, further work remains in standardizing SEP measures across studies and evaluating consistency across domains, tissue types, and time periods. Meta-analyses of epigenetic associations with SEP are offered as one approach to confirm the replication of DNAm loci across studies. The development of DNAm biomarkers of SEP would provide a method for examining its impact on health outcomes in a more robust way, increasing the rigor of epidemiological studies.

Factors affecting HPV infection in U.S. and Beijing females: A modeling study

Background

Human papillomavirus (HPV) infection is an important carcinogenic infection highly prevalent among many populations. However, independent influencing factors and predictive models for HPV infection in both U.S. and Beijing females are rarely confirmed. In this study, our first objective was to explore the overlapping HPV infection-related factors in U.S. and Beijing females. Secondly, we aimed to develop an R package for identifying the top-performing prediction models and build the predictive models for HPV infection using this R package.

Methods

This cross-sectional study used data from the 2009-2016 NHANES (a national population-based study) and the 2019 data on Beijing female union workers from various industries. Prevalence, potential influencing factors, and predictive models for HPV infection in both cohorts were explored.

Results

There were 2,259 (NHANES cohort, age: 20-59 years) and 1,593 (Beijing female cohort, age: 20-70 years) participants included in analyses. The HPV infection rate of U.S. NHANES and Beijing females were, respectively 45.73 and 8.22%. The number of male sex partners, marital status, and history of HPV infection were the predominant factors that influenced HPV infection in both NHANES and Beijing female cohorts. However, condom application was not an independent influencing factor for HPV infection in both cohorts. R package Modelbest was established. The nomogram developed based on Modelbest package showed better performance than the nomogram which only included significant factors in multivariate regression analysis.

Conclusion

Collectively, despite the widespread availability of HPV vaccines, HPV infection is still prevalent. Compared with condom promotion, avoidance of multiple sexual partners seems to be more effective for preventing HPV infection. Nomograms developed based on Modelbest can provide improved personalized risk assessment for HPV infection. Our R package Modelbest has potential to be a powerful tool for future predictive model studies.

Hypochloremia as a novel adverse prognostic factor in acute liver failure

BACKGROUND AND AIMS

Emerging evidence has identified hypochloremia as an independent predictor for mortality in multiple conditions including cirrhosis. Acute liver failure (ALF) is frequently complicated by electrolyte abnormalities. We investigated the prognostic value of hypochloremia in a large cohort of ALF patients from North America.

METHODS

The Acute Liver Failure Study Group (ALFSG) registry is a longitudinal cohort study involving 2588 ALF patients enrolled prospectively from 32 North American academic centres. The primary outcome was a composite of 21-day all-cause mortality or requirement for liver transplantation (death/LT).

RESULTS

Patients with hypochloremia (<98 mEq/L) had a significantly higher 21-day mortality rate (42.1%) compared with those with normal (27.5%) or high (>107 mEq/L) chloride (28.0%) (p < .001). There was lower transplant-free cumulative survival in the hypochloremic group than in the normo- or hyper-chloremic groups (log-rank, χ2 24.2, p < .001). Serum chloride was inversely associated with the hazard of 21-day death/LT with multivariable adjustment for known prognostic factors (adjusted hazard ratio [aHR]: 0.977; 95% CI: 0.969-0.985; p < .001). Adding chloride to the ALFSG Prognostic Index more accurately predicted risk of death/LT in 19% of patients (net reclassification improvement [NRI] = 0.19, 95% CI: 0.13-0.25) but underestimated the probability of transplant-free survival in 34% of patients (NRI = -0.34, 95% CI: -0.39 to -0.28).

CONCLUSIONS

Hypochloremia is a novel independent adverse prognostic factor in ALF. A new ALFSG-Cl Prognostic Index may improve the sensitivity to identify patients at risk for death without LT.

Derivation and Validation a Risk Model for Acute Kidney Injury and Subsequent Adverse Events After Cardiac Surgery: A Multicenter Cohort Study

Purpose

To establish a risk model for acute kidney injury and subsequent adverse events in Chinese cardiac patients.

Patients and Methods

This study included 11,740 patients who had cardiac surgery at 14 institutions in China. Patients were randomly assigned to a derivation cohort (n = 8197) or a validation cohort (n = 3543). Variables ascertained during hospitalization were screened using least absolute shrinkage and selection operator and logistic regression to construct a nomogram model. Model performance was evaluated using C-statistic, calibration curve, and Brier score. The nomogram was further compared with the five conventional models: Mehta score, Ng score, AKICS score, SRI score, and Cleveland Clinic score. Acute kidney injury was defined according to the Kidney Disease Improving Global Outcomes criteria. Subsequent adverse events included mid-term outcomes: death from all causes and major adverse kidney events (defined as composite outcome of death from renal failure, dialysis, and advanced chronic kidney disease).

Results

Acute kidney injury occurred in 3237 (27.6%) patients. The model included 12 predictors. The total score generated from the nomogram ranged from 0 to 556. The nomogram achieved a C-statistic of 0.825 and 0.804 in the derivation and validation cohorts, respectively, and had well-fitted calibration curves. The model performance of the nomogram was better than other five conventional models. After risk stratification, moderate-risk or high-risk groups were associated with significantly higher rates of death from all causes and major adverse kidney events compared with low-risk group during 7-year follow-up.

Conclusion

The nomogram could serve as an effective tool for predicting acute kidney injury and evaluating its subsequent adverse events after cardiac surgery.

Development, validation and clinical utility of a risk prediction model for adverse pregnancy outcomes in women with gestational diabetes: The PeRSonal GDM model

BACKGROUND

The ability to calculate the absolute risk of adverse pregnancy outcomes for an individual woman with gestational diabetes mellitus (GDM) would allow preventative and therapeutic interventions to be delivered to women at high-risk, sparing women at low-risk from unnecessary care. We aimed to develop, validate and evaluate the clinical utility of a prediction model for adverse pregnancy outcomes in women with GDM.

METHODS

A prediction model development and validation study was conducted on data from a observational cohort. Participants included all women with GDM from three metropolitan tertiary teaching hospitals in Melbourne, Australia. The development cohort comprised those who delivered between 1 July 2017 to 30 June 2018 and the validation cohort those who delivered between 1 July 2018 to 31 December 2018. The main outcome was a composite of critically important maternal and perinatal complications (hypertensive disorders of pregnancy, large-for-gestational age neonate, neonatal hypoglycaemia requiring intravenous therapy, shoulder dystocia, perinatal death, neonatal bone fracture and nerve palsy). Model performance was measured in terms of discrimination and calibration and clinical utility evaluated using decision curve analysis.

FINDINGS

The final PeRSonal (Prediction for Risk Stratified care for women with GDM) model included body mass index, maternal age, fasting and 1-hour glucose values (75-g oral glucose tolerance test), gestational age at GDM diagnosis, Southern and Central Asian ethnicity, East Asian ethnicity, nulliparity, past delivery of an large-for-gestational age neonate, past pre-eclampsia, GWG until GDM diagnosis, and family history of diabetes. The composite adverse pregnancy outcome occurred in 27% (476/1747) of women in the development (1747 women) and in 26% (244/955) in the validation (955 women) cohorts. The model showed excellent calibration with slope of 0.99 (95% CI 0.75 to 1.23) and acceptable discrimination (c-statistic 0.68; 95% CI 0.64 to 0.72) when temporally validated. Decision curve analysis demonstrated that the model was useful across a range of predicted probability thresholds between 0.15 and 0.85 for adverse pregnancy outcomes compared to the alternatives of managing all women with GDM as if they will or will not have an adverse pregnancy outcome.

INTERPRETATION

The PeRSonal GDM model comprising of routinely available clinical data shows compelling performance, is transportable across time, and has clinical utility across a range of predicted probabilities. Further external validation of the model to a more disparate population is now needed to assess the generalisability to different centres, community based care and low resource settings, other healthcare systems and to different GDM diagnostic criteria.

FUNDING

This work is supported by the Mothers and Gestational Diabetes in Australia 2 NHMRC funded project #1170847.