The invention provides an
abdominal aortic aneurysm early prediction model construction method,
system and device and a medium. According to the method, a balanced research
queue is constructed through fusion of multi-
modal data (ultrasonic images, biochemical indexes and lipidomics) and tendency
score matching, so that the accuracy of early
risk identification of the
abdominal aortic aneurysm and the model robustness are remarkably improved; furthermore, on the basis of a risk migration strategy, the discrimination capability of the complex
deep learning model is migrated to a simplified
logistic regression model which only needs a simplest lipid
feature set and basic demographic information, so that high prediction performance is maintained, model lightweight, parameter immobilization and clinical
interpretability are realized, the detection cost and the operation complexity are greatly reduced, and the detection efficiency is improved. A high-precision risk prediction technology can be conveniently deployed in a basic medical scene, and finally an
abdominal aortic aneurysm early screening solution with high precision, high
interpretability and wide applicability is formed.