Taiwan’s NTUH develops predictive model for cardiovascular risks

Two databases were utilised with more than 10,000 patients and over 10 years of tracking.

A team from the National Taiwan University Hospital’s (NTUH) Department of Cardiology has developed a risk prediction model for major cardiovascular events including myocardial infarction, stroke, and sudden death.

According to the hospital, the model was created by modifying the US risk prediction model and then continued its development using intelligent analysis.

The team used the Taiwan Cardiovascular Disease Clinical Trial Alliance's two multicentre observational registry databases with more than 10,000 patients and over 10 years of tracking to conduct the research.

“The results showed that in binary classification analysis, intelligent analysis XGBoost and Random forest performed best,” NTUH said.

Meanwhile, a time series validation test was also conducted wherein data showed significant differences in the distribution of different features and event rates. 

“The performance of Random Forest was comparable to that of the logical regression model in terms of numerical performance,” The hospital added.