Heart Rate and Body Temperature Relationship in Children Admitted to PICU: A Machine Learning Approach.

Area of Science:

• Pediatric critical care medicine
• Biomedical informatics
• Clinical physiology

Background:

• Vital signs like body temperature (BT) and heart rate (HR) are critical in pediatric intensive care units (PICUs).
• Limited research exists on the HR-BT relationship in critically ill children.
• Understanding these dynamics is vital for accurate patient assessment.

Purpose of the Study:

• To investigate the relationship between heart rate (HR) and body temperature (BT) in pediatric patients (0-18 years) admitted to a PICU.
• To apply machine learning (ML) techniques to model these complex associations.
• To compare ML model performance against traditional linear assumptions.

Main Methods:

• Utilized Gradient Boosting Machines (GBM) with Quantile Regression (QR) for modeling HR, BT, and age.
• Employed hyperparameter tuning to optimize model performance.
• Analyzed data from 4006 pediatric patients admitted to CHU Sainte-Justine (CHUSJ) Hospital.

Main Results:

• Observed decreasing HR with increasing age and increasing HR with higher BT ranges.
• Linear models inaccurately estimated HR, particularly in younger children across different BT ranges.
• The GBM model provided improved accuracy for HR prediction, considering age and BT percentiles.

Conclusions:

• The relationship between HR, BT, and age in critically ill children is non-linear.
• Machine learning offers a more accurate approach to modeling these vital sign dynamics.
• Findings challenge traditional linear assumptions and inform clinical decision-making in PICUs.