Predicting Risk of Mortality in Pediatric ICU Based on Ensemble Step-Wise Feature Selection

Abstract

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 data-driven 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.

simpleNomo: A Python Package of Making Nomograms for Visualizable Calculation of Logistic Regression Models

“…In this study, we provided examples of constructing nomograms based on several published logistic regression models that do not currently include nomograms. Specifically, we selected the evolution of low back pain patients prediction model [ 17 ], the Epstein-Barr virus reactivation (EBV) prediction model [ 18 ], the refractory/recurrent CMV infection prediction model [ 16 ], the severe acute graft-versus-host disease (aGVHD) prediction model [ 19 ], and the risk of mortality in pediatric intensive care unit (ICU) prediction model [ 20 ]. To construct the nomograms, we followed a sequential process of retrieving logistic regression model information from the published papers, constructing the meta-data file, and developing the nomogram using either the simpleNomo software or the online generator.…”

“… Develop nomograms from established Risk of Mortality in Pediatric ICU prediction model [ 20 ]. (A) The original presentation of logistic regression model information [ 20 ].…”

“… Develop nomograms from established Risk of Mortality in Pediatric ICU prediction model [ 20 ]. (A) The original presentation of logistic regression model information [ 20 ]. (B) The excel containing the logistic regression model information for generating nomogram.…”

Machine Learning-Based Systems for the Anticipation of Adverse Events After Pediatric Cardiac Surgery

“…Most of the proposed models were developed to make one prediction per patient encounter using data within the first hours after ICU admission (37)(38)(39)(40), rather than predict the risk of mortality continuously across the entire encounter. However, in order to continually assess individual patient's risk of clinical deterioration or mortality it is important to integrate information not only from a single time point, as the current scoring systems do, but also data from previous time points, that is, longitudinal temporal data.…”

Machine learning-based prediction of in-hospital mortality for critically ill patients with sepsis-associated acute kidney injury

“…Model construction techniques of machine learning methods, which include random forest (RF), extreme gradient boosting (XGBoost) and other methods, have been widely used in the medical field [ 11 , 12 ]. Integrating feature ranking and screening predictors step by step and obtaining a subset of valid features were also helpful for improving the discrimination and accuracy of a prediction model [ 13 ]. The new variable screening methods in combination with multiple machine learning techniques may further increase modeling effectiveness.…”