Prediction and causal inference of cardiovascular and cerebrovascular diseases based on lifestyle questionnaires

Area of Science:

• Cardiology and Neurology
• Medical Informatics
• Public Health

Background:

• Cardiovascular and cerebrovascular diseases (CCVD) are leading causes of mortality in Japan.
• Effective preventative measures, early diagnosis, and treatment are crucial for mitigating CCVD impact.
• A diagnostic model was developed using specific health examination data to identify patients with ischemic heart disease (IHD), stroke, or both.

Purpose of the Study:

• To develop and validate a diagnostic model for identifying patients at risk of IHD, stroke, or CCVD.
• To analyze lifestyle habits influencing CCVD development using causal inference methods.
• To improve early detection and resource allocation for CCVD screening.

Main Methods:

• Utilized health examination data from 473,734 individuals (≥40 years) in Kanazawa, Japan (2009-2018).
• Employed four machine learning algorithms: Random Forest, Logistic Regression, Light Gradient Boosting Machine, and XGBoost.
• Applied five causal inference methods to analyze lifestyle factors and their relationship with CCVD.

Main Results:

• The XGBoost model demonstrated superior performance with high AUC values for stroke (0.770), IHD (0.758), and CCVD (0.845).
• Causal inference identified a link between reduced mastication and weight gain.
• Lifestyle behavior changes were observed post-CCVD onset.

Conclusions:

• The developed prediction algorithm can effectively screen for asymptomatic myocardial ischemia and stroke.
• This tool enables efficient identification of high-risk patients for targeted secondary testing.
• The findings support improved resource allocation and secondary prevention strategies for CCVD.