Objective: To develop an illness severity score that predicts short-term mortality, based on a small number of readily available measurements, and overcomes limitations of the SOFA score, for use in research involving...Objective: To develop an illness severity score that predicts short-term mortality, based on a small number of readily available measurements, and overcomes limitations of the SOFA score, for use in research involving large-scale electronic health records. Design: Retrospective analysis of electronic records for 37,739 adult inpatients. Setting: A single tertiary care hospital system from 2016-2022. Patients: 37,739 adult ICU patients. Interventions: IMPS was developed using logistic regression with the 6 SOFA components, age, sex and missingness indicators as predictors, and 10-day mortality as the outcome. This was compared with SOFA with median imputation. Measurements and Main Results: Discrimination was evaluated by AUROC, calibration by comparing predicted and observed mortality. IMPS showed excellent discrimination (AUROC 0.80) and calibration. It outperformed SOFA alone (AUROC 0.70) and with age/sex (0.74). Conclusions: By retaining continuous data, adding age, allowing for missingness, and optimizing weights based on empirical mortality association, IMPS achieved substantially better mortality prediction than the original SOFA.展开更多
Background.Prediction of mortality risk in intensive care units(ICU)is an important task.Data-driven methods such as scoring systems,machine learning methods,and deep learning methods have been investigated for a long...Background.Prediction of mortality risk in intensive care units(ICU)is an important task.Data-driven methods such as scoring systems,machine learning methods,and deep learning methods have been investigated for a long time.However,few datadriven methods are specially developed for pediatric ICU.In this paper,we aim to amend this gap—build a simple yet effective linear machine learning model from a number of hand-crafted features for mortality prediction in pediatric ICU.Methods.We use a recently released publicly available pediatric ICU dataset named pediatric intensive care(PIC)from Children’s Hospital of Zhejiang University School of Medicine in China.Unlike previous sophisticated machine learning methods,we want our method to keep simple that can be easily understood by clinical staffs.Thus,an ensemble step-wise feature ranking and selection method is proposed to select a small subset of effective features from the entire feature set.A logistic regression classifier is built upon selected features for mortality prediction.Results.The final predictive linear model with 11 features achieves a 0.7531 ROC-AUC score on the hold-out test set,which is comparable with a logistic regression classifier using all 397 features(0.7610 ROC-AUC score)and is higher than the existing well known pediatric mortality risk scorer PRISM III(0.6895 ROC-AUC score).Conclusions.Our method improves feature ranking and selection by utilizing an ensemble method while keeping a simple linear form of the predictive model and therefore achieves better generalizability and performance on mortality prediction in pediatric ICU.展开更多
文摘Objective: To develop an illness severity score that predicts short-term mortality, based on a small number of readily available measurements, and overcomes limitations of the SOFA score, for use in research involving large-scale electronic health records. Design: Retrospective analysis of electronic records for 37,739 adult inpatients. Setting: A single tertiary care hospital system from 2016-2022. Patients: 37,739 adult ICU patients. Interventions: IMPS was developed using logistic regression with the 6 SOFA components, age, sex and missingness indicators as predictors, and 10-day mortality as the outcome. This was compared with SOFA with median imputation. Measurements and Main Results: Discrimination was evaluated by AUROC, calibration by comparing predicted and observed mortality. IMPS showed excellent discrimination (AUROC 0.80) and calibration. It outperformed SOFA alone (AUROC 0.70) and with age/sex (0.74). Conclusions: By retaining continuous data, adding age, allowing for missingness, and optimizing weights based on empirical mortality association, IMPS achieved substantially better mortality prediction than the original SOFA.
文摘Background.Prediction of mortality risk in intensive care units(ICU)is an important task.Data-driven methods such as scoring systems,machine learning methods,and deep learning methods have been investigated for a long time.However,few datadriven methods are specially developed for pediatric ICU.In this paper,we aim to amend this gap—build a simple yet effective linear machine learning model from a number of hand-crafted features for mortality prediction in pediatric ICU.Methods.We use a recently released publicly available pediatric ICU dataset named pediatric intensive care(PIC)from Children’s Hospital of Zhejiang University School of Medicine in China.Unlike previous sophisticated machine learning methods,we want our method to keep simple that can be easily understood by clinical staffs.Thus,an ensemble step-wise feature ranking and selection method is proposed to select a small subset of effective features from the entire feature set.A logistic regression classifier is built upon selected features for mortality prediction.Results.The final predictive linear model with 11 features achieves a 0.7531 ROC-AUC score on the hold-out test set,which is comparable with a logistic regression classifier using all 397 features(0.7610 ROC-AUC score)and is higher than the existing well known pediatric mortality risk scorer PRISM III(0.6895 ROC-AUC score).Conclusions.Our method improves feature ranking and selection by utilizing an ensemble method while keeping a simple linear form of the predictive model and therefore achieves better generalizability and performance on mortality prediction in pediatric ICU.
