The present invention discloses a method and
system for constructing a prediction model for postoperative
cognitive decline in elderly patients, including: generating a preoperative
feature matrix; using a static risk prediction model to evaluate the initial
risk probability of POCD; introducing a variational
autoencoder to model the preoperative
feature matrix, and generating a personalized preoperative cognitive
state distribution of the patient by learning the implicit distribution of the preoperative
feature matrix, thereby completing the initial construction of a digital twin; dynamically collecting cognitive scale scores, physiological indicators, voice features, and behavioral activity data at different time points after
surgery to generate a
time series feature matrix; inputting the preoperative feature matrix and the
time series feature matrix into a dynamic
time series model to update the patient's POCD
risk probability and cognitive
state distribution; simulating a personalized cognitive
recovery trajectory, predicting the time it takes for the patient to return to the baseline state, and outputting the optimal intervention strategy to reduce the risk of POCD. The present invention can fine-tune parameters in real time according to newly added data after
surgery, provide rapidly updated prediction results, thereby overcoming the problem that existing models are difficult to update prediction results in real time.