Improving Cardiovascular Disease Prediction With Machine Learning Using Mental Health Data: A Prospective UK Biobank...

Area of Science:

• Computational biology and bioinformatics
• Cardiovascular medicine
• Psychiatry and psychology

Background:

• Cardiovascular disease (CVD) risk prediction is crucial for timely patient intervention.
• The link between mental health and CVD is established, yet often ignored in predictive models.
• Current CVD models lack comprehensive psychological data and are trained on limited populations.

Purpose of the Study:

• To evaluate the effectiveness of incorporating psychological data into a machine learning (ML) model for improved CVD prediction.
• To assess the impact of psychological factors on the accuracy of CVD risk assessment.

Main Methods:

• Utilized a large UK Biobank dataset (n=375,145) to analyze CVD correlations with traditional and psychological risk factors.
• Developed an ensemble ML model comprising five algorithms: decision tree, random forest, XGBoost, support vector machine, and deep neural network.
• Trained and tested the model using two datasets: one with traditional risk factors only, and another including both traditional and psychological factors.

Main Results:

• The ensemble ML model achieved 71.31% accuracy in predicting CVD using traditional risk factors alone.
• Incorporating psychological factors increased the model's prediction accuracy to 85.13%.
• The ensemble ML model demonstrated superior accuracy and robustness compared to its individual constituent algorithms.

Conclusions:

• Integrating mental health assessment data into ensemble ML models substantially enhances CVD prediction accuracy.
• This approach offers a more robust and accurate method for CVD risk prediction than models relying solely on traditional factors.