Research On Prevention and Control Mode and Governance Mechanism of Chronic Diseases Based on Medical Statistics

Abstract

Faced with the triple challenges of high incidence, common comorbidity and heavy economic burden of chronic diseases in China, the traditional passive mode of "hospital-centered" is unsustainable. This study puts forward a research framework of "medical statistical data mining-dynamic prevention and control model-multi-collaborative governance", and systematically answers three scientific questions: "how to extract key features, how to achieve accurate stratification, and how to ensure sustainable governance". Based on more than 45,000 longitudinal queues of multi-center HIS/LIS/EHR, wearable devices and environmental monitoring, cross-domain data fusion is completed by using differential privacy (ε=1.5) and federated learning; Combined with random forest and XGBoost, the risk prediction model was established (AUC=0.82), and five heterogeneous groups were excavated by Apriori and K-means (35.7% in the elderly comorbid disability group and 18.5% in the environmentally sensitive hypertension group). Based on this, a data-driven prevention and control model of "accurate prediction-graded intervention-effect closed loop" is designed, and the long-term cost-effectiveness of 12 governance strategies is quantitatively compared on PEST-SD simulation platform. Further build a blockchain data sharing and performance compensation mechanism of "government-led, departmental collaboration, social participation and personal responsibility" to realize the governance paradigm transformation from "one-way decision-making by the government" to "multi-subject collaboration" and from "intervention after the event" to "full management". The research results provide a replicable Chinese plan for the reform of health insurance per head, the construction of healthy cities and the implementation of healthy China strategy.