A gender specific risk assessment of coronary heart disease based on physical examination data

Area of Science:

• Cardiovascular Disease Research
• Biomedical Informatics
• Public Health Screening

Background:

• Coronary heart disease (CHD) prevention requires effective large-scale risk screening.
• Physical examination data offers a practical, comprehensive resource for health assessments.
• Existing risk assessment tools may lack gender specificity or ease of clinical application.

Purpose of the Study:

• To develop and validate a gender-specific system for coronary heart disease (CHD) risk assessment using physical examination data.
• To create a user-friendly CHD risk model and scorecard for widespread clinical and personal use.
• To establish an accessible online platform for CHD risk evaluation and lifestyle management.

Main Methods:

• A large dataset of 39,538 CHD patients and 640,465 healthy individuals was utilized.
• Feature selection identified ten key risk factors from fifty physical examination characteristics.
• A fully connected network (FCN) model was developed for CHD risk prediction, alongside a logistic regression (LR) based scorecard.

Main Results:

• The FCN model demonstrated high performance with AUCs ranging from 0.8671 to 0.9006 across male and female cohorts on independent and external validation sets.
• The CHD risk scorecard showed only a minor decrease in AUC (<0.05) compared to the FCN model, indicating minimal information loss.
• An online CHD risk assessment system was successfully established and made publicly available.

Conclusions:

• The developed gender-specific system provides an accurate and efficient method for large-scale coronary heart disease risk screening.
• The CHD risk scorecard offers a practical, low-information-loss alternative for clinical implementation.
• The online assessment system promotes personalized lifestyle management for CHD prevention.