Joint team develops AI-based cardiovascular disease prediction model
A joint research team has developed an artificial intelligence (AI) -based cardiovascular disease prediction model for Koreans.
The team, led by Professor Kang Si-hyuk from Bundang Hospital at Seoul National University and Professor Cho Sang-young from Hospital at Gyeongsang Changwon National University, used data from around 220,000 adults aged 40 to 80 who participated in the health check of the national health insurance service from 2009 to 2010.
The predictive model predicts the risk of cardiovascular disease using subject data, such as age, gender, systolic blood pressure, cholesterol, smoking, and history of diabetes.
The researchers confirmed that 7,819, or 3.51 percent of the total 220,000 people, had atherosclerotic cardiovascular disease during a five-year follow-up.
After analyzing the accuracy of their model for predicting cardiovascular disease risk, the team confirmed that existing models generally show a prediction accuracy of 70-80%. Specifically, the prediction accuracy by the pooled cohort equation of the US AHA prediction model, which was the primary comparison target, was 73.8%.
In comparison, the model developed by the team showed a slightly better accuracy of 75.1%, confirming its superior predictive performance compared to the existing model.
“The benefit of machine learning is increased accuracy and reduced errors even with the same settings,” the team said. “We anticipate that the new model will be advantageous for individualized and personalized treatments for patients because it can accurately calculate individual risk.”
Professor Kang said, “AI machine learning will continue to be used not only for health, but also to enrich life.”
Suppose the researchers expand the application of machine learning in the medical field. In this case, it will be possible to accurately predict the risk of diseases and offer more effective treatments while reducing human effort, he added.
Professor Cho also noted that the heart of the study was to select the risk group through the cardiovascular disease prediction model and to present an effective preventive treatment policy.
“As the study confirmed that the predictive power of our developed model is superior compared to the model previously used, we will continue to research the development and use of a high precision assessment tool,” he said. he declares.