One Year Risk Prediction Model for Cardiovascular Disease for Adults with Asthma in England

Area of Science:

• Cardiovascular health
• Asthma management
• Machine learning in medicine

Background:

• Cardiovascular disease (CVD) poses a global health challenge.
• Predicting CVD risk in asthma patients using machine learning (ML) is underexplored.

Purpose of the Study:

• To develop and evaluate ML models for CVD risk prediction in asthma patients.
• To identify the most accurate and practical model for clinical implementation.

Main Methods:

• A cohort study of 641,042 participants using electronic healthcare records.
• Exploration of various ML algorithms: logistic regression, penalized logistic regression, decision trees, random forest, and gradient boost.

Main Results:

• Penalized logistic regression showed the best discriminatory power (AUC = 0.85).
• Gradient boost model offered the best calibration.
• Previous cardiovascular events, age, and cardiovascular medication prescriptions were key predictors.

Conclusions:

• A novel prediction model for 1-year CVD risk post-asthma diagnosis was developed.
• Penalized logistic regression is a suitable, transparent model for screening low-risk patients.
• ML models outperformed traditional risk prediction methods, potentially reducing unnecessary treatments by 52%.