A Technical Performance Study and Proposed Systematic and Comprehensive Evaluation of an ML-based CDS Solution for...

Area of Science:

• Pediatric Pulmonology
• Clinical Informatics
• Machine Learning in Healthcare

Background:

• Optimal pediatric asthma care requires efficient access to scattered patient information.
• Clinicians face challenges consolidating data from structured and unstructured clinical notes.
• Current workflows lack adequate tools to support data integration for asthma management.

Purpose of the Study:

• To develop a machine learning-based clinical decision support (CDS) system for pediatric asthma care.
• To integrate a predictive model for asthma exacerbation risk into clinical workflows.
• To enhance clinical usability through contextual data, supporting information, and model transparency.

Main Methods:

• Development of a machine learning model to predict pediatric asthma exacerbation risk.
• Integration of the model into a clinical workflow emphasizing usability.
• Focus on contextual data, supporting information, and model explainability.

Main Results:

• The asthma exacerbation prediction model achieved an AUC-ROC of 0.8.
• The developed CDS system aims to alleviate the burden on clinicians for data consolidation.
• The informatics-based process prioritizes clinical practice needs.

Conclusions:

• Machine learning technology can effectively support pediatric asthma care.
• The proposed CDS system demonstrates potential for improving clinical decision-making.
• Integrating predictive models enhances efficiency and accessibility of patient information for better asthma management.