Investigating Machine Learning Techniques for Predicting Risk of Asthma Exacerbations: A Systematic Review.

Area of Science:

• Medical Informatics
• Artificial Intelligence in Healthcare
• Respiratory Medicine

Background:

• Asthma affects over 200 million globally, causing significant mortality.
• Machine learning (ML) is increasingly used in healthcare for decision support.
• Predicting asthma exacerbations with ML remains an area with uncertainties.

Purpose of the Study:

• To systematically review recent ML applications for predicting asthma attack risk.
• To aid in asthma control and management through improved exacerbation prediction.
• To propose a conceptual model for leveraging ML in early asthma attack detection.

Main Methods:

• Systematic review of 20 studies published between January 2010 and February 2023.
• Initial search identified 860 studies across five databases.
• Included studies utilized various data sources for ML-based asthma risk prediction.

Main Results:

• Reviewed studies employed ML techniques for asthma risk prediction using diverse data (clinical, biological, environmental, etc.).
• Approaches varied, with some studies predicting risk category and others predicting exacerbation probability.
• A conceptual model was proposed to integrate ML and data sources for early asthma attack detection.

Conclusions:

• ML offers potential for enhanced asthma exacerbation prediction.
• Diverse data sources are crucial for developing effective ML models.
• Further research is needed to refine ML applications for proactive asthma management.