Approved Research

The Prediction of Cardiovascular Disease Based on Graph Nerual Network

Lay summary

Cardiovascular diseases (CVDs) are a group of disorders that affect the heart and blood vessels. CVD generally develops without any symptoms at first and many people may be unaware they have it until a life-threatening event occurs, such as a heart attack or a stroke.

The aim of our research is to develop better tools to diagnose CVD at an early stage, so that we can identify individuals in need of treatment before a life-threatening event occurs. In our research, we want to enable multi-modal data such as excel data, text data and image data. We are interested in every factors which would cause cardiovascular diseases explicitly or implicitly, generally it includes time-series data such as physical examinations and analyses of blood samples during a period, and static data, like family history of heredity, age, height, weight, lifestyle and environment.

There are lots of biomarkers, so we first identify the best combination of biomarkers, which then be selected and evaluated using advanced statistical methods. After that, we would use a data driven mehod llike machine learning,  to predict the risk of CVD before life-threatening event occurs.

CVD is today the most common death cause worldwide. If we can propose a powerful and effective method and demo, we can help doctors and patients to diagnose if their patients or themselves suffer from CVD easily and conveniently. And we plan to finish the project in 2 years.