Related Experiment Video
Updated: Jul 20, 2025
Surfactant Depletion Combined with Injurious Ventilation Results in a Reproducible Model of the Acute Respiratory Distress Syndrome ARDS
Published on: April 7, 2021
A data-driven model for early prediction of need for invasive mechanical ventilation in pediatric intensive care unit
Sanjukta N Bose1, Andrew Defante2, Joseph L Greenstein3
1Enterprise Data and Analytics, University of Maryland Medical System, Linthicum Heights, MD, United States of America.
Insights
This study developed a predictive model to identify pediatric intensive care unit (PICU) patients needing mechanical ventilation (MV) early. The model, using clinical and medication data, provides a crucial early warning period to potentially avert MV.
Area of Science:
- Pediatric Critical Care Medicine
- Medical Informatics
- Machine Learning in Healthcare
Background:
- Acute respiratory failure is a critical condition in pediatric intensive care units (PICUs).
- Mechanical ventilation (MV) is often required for survival in severe cases.
- Early recognition of patients at risk for MV can improve clinical outcomes.
Purpose of the Study:
- To develop a data-driven model for the early prediction of MV necessity in PICU patients.
- To establish an early warning period for clinicians to intervene before MV is required.
Main Methods:
- A retrospective observational study of 13,651 PICU patients.
- Development of a prediction model using XGBoost and a convolutional neural network (CNN) with medication history.
- Calculation of a time-varying risk-score and an optimal threshold from ROC curves to determine the early prediction point (EPP) and early warning time (EWT).
- Spectral clustering to identify patient groups based on risk-score trajectories.
Main Results:
- The clinical and medication history-based model achieved an AUROC of 0.89.
- The model demonstrated high specificity (0.95) and NPV (0.95), with a median EWT of 9.9 hours.
- Clustering identified three patient groups post-EPP, with the highest risk group showing a PPV of 0.92.
Conclusions:
- This study presents a novel method utilizing medication history for predicting MV needs in pediatric patients.
- The developed model provides a valuable early warning period, potentially enabling timely interventions.
- The findings highlight the utility of machine learning in proactive critical care management.
Rationale:
Acute respiratory failure is a life-threatening clinical outcome in critically ill pediatric patients. In severe cases, patients can require mechanical ventilation (MV) for survival. Early recognition of these patients can potentially help clinicians alter the clinical course and lead to improved outcomes.
Objectives:
To build a data-driven model for early prediction of the need for mechanical ventilation in pediatric intensive care unit (PICU) patients.
Methods:
The study consists of a single-center retrospective observational study on a cohort of 13,651 PICU patients admitted between 1/01/2010 and 5/15/2018 with a prevalence of 8.06% for MV due to respiratory failure. XGBoost (extreme gradient boosting) and a convolutional neural network (CNN) using medication history were used to develop a prediction model that could yield a time-varying "risk-score"-a continuous probability of whether a patient will receive MV-and an ideal global threshold was calculated from the receiver operating characteristics (ROC) curve. The early prediction point (EPP) was the first time the risk-score surpassed the optimal threshold, and the interval between the EPP and the start of the MV was the early warning period (EWT). Spectral clustering identified patient groups based on risk-score trajectories after EPP.
Results:
A clinical and medication history-based model achieved a 0.89 area under the ROC curve (AUROC), 0.6 sensitivity, 0.95 specificity, 0.55 positive predictive value (PPV), and 0.95 negative predictive value (NPV). Early warning time (EWT) median [inter-quartile range] of this model was 9.9[4.2-69.2] hours. Clustering risk-score trajectories within a six-hour window after the early prediction point (EPP) established three patient groups, with the highest risk group's PPV being 0.92.
Conclusions:
This study uses a unique method to extract and apply medication history information, such as time-varying variables, to identify patients who may need mechanical ventilation for respiratory failure and provide an early warning period to avert it.
Related Concept Videos
01:23Mechanical Ventilation II: Invasive Ventilation
Negative-Pressure Ventilators
Negative-pressure ventilators create a vacuum around the chest or body to draw air into the lungs, simulating breathing. This method does not require an...
01:29Mechanical Ventilation I: Indication and Settings
01:23Mechanical Ventilation III: Noninvasive Ventilation
Noninvasive Positive-Pressure Ventilation...
01:20Assessment of Ventilation I: Respiratory Rate
A Ventilation assessment is critical for monitoring a patient's health status. Respiration, one of the most accessible vital signs, provides insights into the function of numerous body systems and can indicate serious health issues, such as brainstem injuries from head trauma.
Critical Guidelines for Assessing Ventilation:
01:14Ventilatory Modes
There are three ventilatory modes: full support, partial support, and spontaneous. These are described below.
Full Support Modes
Full support modes include controlled mechanical ventilation, continuous mandatory...