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

Abstract

Vital signs are crucial clinical measures, with body temperature (BT) and heart rate (HR) being particularly significant. While their association has been studied in adults and children, research in Pediatric Intensive Care Unit (PICU) settings remains limited despite the critical conditions of these patients.

Objective: This study examines the relationship between HR and BT in children aged 0 to 18 admitted to the PICU at CHU Sainte-Justine (CHUSJ) Hospital.

Methods: Machine learning (ML) techniques, including Gradient Boosting Machines (GBM) with Quantile Regression (QR), were applied to capture the relationship between HR, BT, and age, optimizing model performance through hyperparameter tuning.

Results: Analyzing data from 4006 children, we observed a consistent trend of decreasing HR with increasing age and rising HR with higher BT ranges. Linear models often underestimated HR at lower BT ranges and overestimated it at higher ranges, especially in younger age groups. The GBM model demonstrated improved accuracy and supported a user-friendly interface for HR predictions based on BT, age, and HR percentiles. Qualitative observations indicated that linear models underestimated HR at lower BT ranges and overestimated it at higher ones, particularly in younger children. These findings challenge the direct linear association assumed in prior studies.

Conclusion: This study provides new insights into the non-linear dynamics between HR, BT, and age in critically ill children, emphasizing further research to quantify and understand these relationships.

Significance: By refining predictive models and re-evaluating traditional assumptions, this work provides valuable insights for improving clinical decision-making in PICU settings.