Identifying Risk Indicators of Cardiovascular Disease in Fasa Cohort Study (FACS): An Application of Generalized...

Area of Science:

• Epidemiology
• Biostatistics
• Public Health

Background:

• Cardiovascular disease (CVD) is a leading global cause of mortality.
• Accurate prediction of CVD is crucial for public health initiatives.
• Early detection frameworks are needed to mitigate CVD impact.

Purpose of the Study:

• To predict cardiovascular disease (CVD) using key indicators.
• To develop and evaluate a tree-based statistical framework for CVD detection.
• To compare the predictive performance of different statistical models for CVD.

Main Methods:

• Utilized baseline data from the Fasa Cohort Study (FACS).
• Fitted ordinary Tree, generalized linear mixed models (GLMM), and GLMM tree models.
• Compared predictive power using metrics like Area Under the Curve (AUC) and specificity/sensitivity.

Main Results:

• Analyzed data from 9499 participants aged 35-70 years.
• Identified age, cholesterol, smoking, glucose, and family history as significant CVD predictors.
• GLMM achieved the highest AUC (0.81), closely followed by the GLMM tree (0.80), outperforming the ordinary tree (0.58).

Conclusions:

• The GLMM tree offers a simpler interpretation and fewer assumptions compared to standard tree models.
• The GLMM tree demonstrates comparable performance to GLMM for CVD prediction.
• Updated statistical models like the GLMM tree can enhance proactive patient detection and planning for CVD.