Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: Apr 22, 2026

Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack

Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack

Published on: May 15, 2020

Machine Learning-Based Prediction of Critical Deterioration in the PICU.

Sanjiv D Mehta^1,2,3, Eamonn Tweedy⁴, Victor M Ruiz⁴

¹Division of Critical Care Medicine, Department of Anesthesiology and Critical Care Medicine, The Children's Hospital of Philadelphia, Philadelphia, PA.

Critical Care Explorations

|April 21, 2026

Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Association of Birth Weight with Delivery Room Endotracheal Intubation Success in Newborn Infants.

The Journal of pediatrics·2026

Same author

Volume-controlled mechanical ventilation during cardiopulmonary resuscitation: A systematic review and meta-analysis.

Resuscitation·2026

Same author

PEAK-II Trach: A Multi-Institutional Quality Improvement/Simulation Study Assessing Pediatric Emergency Readiness.

Otolaryngology--head and neck surgery : official journal of American Academy of Otolaryngology-Head and Neck Surgery·2026

Same author

Early prediction of unplanned critical care transfers in children using EHR-based ensemble machine learning.

JAMIA open·2026

Same author

Variability in Vasoactive Medication use Across Pediatric Intensive Care Units: A PICU Data Collaborative Study, 2010-2022.

Journal of intensive care medicine·2026

Same author

Association of Extubation Timing and Treatment Failure After Successful Extubation Readiness Trial in Pediatric Patients.

Respiratory care·2026

A new machine learning model, P-WIN, accurately predicts critical deterioration events (CDEs) in pediatric intensive care units (PICUs) up to 12 hours in advance. This advanced warning system significantly reduces alert burden, enabling proactive patient care.

Area of Science:

Pediatric critical care medicine
Machine learning applications in healthcare
Clinical decision support systems

Background:

Existing pediatric intensive care unit (PICU) early warning systems lack accuracy and timeliness.
Machine learning (ML) may enhance prediction of critical deterioration events (CDEs).
The operational utility of ML models versus current tools needs clarification.

Purpose of the Study:

To develop an ML model for early CDE detection in PICU patients.
To evaluate the model's operational utility using alert burden analysis.
To compare the ML model's performance against existing PICU warning tools.

Main Methods:

Developed an ensemble of extreme gradient-boosted models (P-WIN) using 550 features.

Keywords:

cardiac arrest prevention clinical deterioration critical care medicine machine learning pediatrics

Related Experiment Videos

Last Updated: Apr 22, 2026

Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack

Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack

Published on: May 15, 2020

Trained P-WIN on historical PICU admission data (2014-2020).

Validated P-WIN on a separate cohort (2021-2022) predicting CDEs at 1-12-hour horizons.

Main Results:

P-WIN achieved high discrimination (AUROC 0.95 at 2 hours, 0.93 at 12 hours).
The model predicted CDEs a median of 10.17 hours in advance.
P-WIN generated one-third the alert burden of existing tools at equivalent sensitivity.

Conclusions:

P-WIN accurately predicts PICU CDEs up to 12 hours prior with a low alert burden.
The model facilitates a shift from reactive rescue to proactive preparation.
P-WIN offers a viable opportunity for improved patient management in PICUs.