Related Experiment Video
Updated: Aug 10, 2026
Implementation of a Real-Time Psychosis Risk Detection and Alerting System Based on Electronic Health Records using CogStack
Published on: May 15, 2020
Validation of a hospital-wide risk stratification model for predicting code events among critically ill children
Colleen M Badke1,2, Sierra Strutz3, Austin Wang4
1Division of Critical Care Medicine, Ann & Robert H. Lurie Children's Hospital of Chicago, Chicago, IL, United States.
Insights
The pediatric Critical event Risk Evaluation and Scoring Tool (pCREST) effectively predicts critical events in children, showing strong performance in early identification of code events in the PICU up to 18 hours prior.
Area of Science:
- Pediatric critical care medicine
- Machine learning in healthcare
- Clinical decision support systems
Background:
- The pediatric Critical event Risk Evaluation and Scoring Tool (pCREST) is a machine learning model designed for continuous risk prediction of deterioration in pediatric patients.
- Early identification of critical events is crucial for improving outcomes in the pediatric intensive care unit (PICU).
Purpose of the Study:
- To externally validate the performance of the pCREST model in the early identification of code events within a PICU setting.
- To assess the predictive capabilities of pCREST at various time points preceding a code event.
Main Methods:
- A single-center, retrospective external validation study was conducted from 2020 to 2023.
- pCREST scores were calculated for patient observations, and predictive performance was evaluated for code events (cardiopulmonary resuscitation and/or code-dose epinephrine) within 12 hours.
- Transfer learning using a recurrent neural network (pCREST-RNN) was explored, but did not enhance performance.
Main Results:
- The study included 5254 patient encounters, with 2.6% experiencing a code event.
- pCREST demonstrated strong predictive performance for code events within 12 hours (c-statistic 0.80).
- The model showed high sensitivity up to 18 hours before an event, with a 13-fold improvement in detecting events over prevalence.
Conclusions:
- The pCREST model is effective in predicting code events in an external PICU setting.
- pCREST provides valuable early warnings for critical events, with high performance demonstrated several hours before the event occurs.
Objective:
The pediatric Critical event Risk Evaluation and Scoring Tool (pCREST) is a validated, machine learning model that continuously predicts the risk of deterioration in children across hospital units. Our objective was to test the performance of pCREST in the early identification of code events in the pediatric intensive care unit (PICU).
Methods:
This was a single-center, retrospective external validation study (2020-2023). The primary outcome was a code (receipt of cardiopulmonary resuscitation and/or code-dose epinephrine) within 12 h of any recorded vital sign or laboratory result. Code events were extracted from our hospital communications platform and electronic health record, and the first code event for each patient was identified. pCREST scores were calculated for each observation and their predictive performance assessed across various time points. Transfer learning was performed via a recurrent neural network (pCREST-RNN).
Results:
Of 5254 encounters in the cohort, 2.6% experienced at least one code event. pCREST demonstrated good performance in predicting code events within 12 h (c-statistic 0.80), and also demonstrated a 13-fold improvement in detecting the outcome over the outcome's prevalence (area under the precision recall curve: 0.034). Sensitivity analysis demonstrated high performance up to 18 h before the event. Transfer learning did not improve performance.
Conclusions:
pCREST performs well at predicting code events in external setting at several time points before the outcome.