This research develops risk prediction models for coronary artery stenosis and vulnerable plaques. The coronary artery stenosis model aims to predict stenosis using multimodal deep learning by integrating text, structured numerical data, and imaging features, focusing on metrics like maximum and cumulative stenosis. The vulnerable plaque model seeks to identify early formation indicators, allowing for timely interventions to prevent plaque rupture, using similar data integration techniques. Additionally, a decision support system is created, comprising a patient database, risk prediction models, and a high-risk alert module. This system facilitates real-time notifications to healthcare providers when risk thresholds are exceeded, enabling personalized treatment planning and improved patient outcomes.

Inclusion Criteria:

Patients who have undergone multiple consecutive CCTA examinations.

Exclusion Criteria:

Patients who have undergone only a single CCTA examination. Patients whose CCTA image quality is poor and cannot be analyzed.

Patients who have undergone multiple consecutive CCTA examinations.