Multi-source heterogeneous cardiovascular disease risk prediction federated learning optimization method and system

By standardizing and differentiating multi-source heterogeneous cardiovascular disease data, and combining secure multi-party computation and manifold learning to optimize the global parameter set, the adaptability and efficiency problems of traditional neural networks in multi-source heterogeneous data processing are solved, and efficient cardiovascular disease risk assessment is achieved.

CN121502320BActive Publication Date: 2026-05-26INNER MONGOLIA MEDICAL UNIV
View PDF 2 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
INNER MONGOLIA MEDICAL UNIV
Filing Date
2025-11-12
Publication Date
2026-05-26

AI Technical Summary

Technical Problem

Traditional neural networks are not adaptable enough when processing multi-source heterogeneous cardiovascular disease data. They have difficulty automatically extracting effective features, have a large computational load, low training efficiency, and the imbalance of data from different sources leads to insufficient model generalization ability.

Method used

We employ a federated learning optimization method for multi-source heterogeneous cardiovascular disease risk prediction. By standardizing feature data, differentially training algorithms, local computation, secure multi-party computation, and manifold learning, we generate an optimized global parameter set and perform risk assessment through a multilayer perceptron network with an attention mechanism.

Benefits of technology

It improves the efficiency of collaborative utilization of multi-source heterogeneous data, ensures data privacy and security, enhances the adaptability and generalization ability of the model, and generates accurate cardiovascular disease risk assessment results.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121502320B_ABST
    Figure CN121502320B_ABST
Patent Text Reader

Abstract

The application provides a multi-source heterogeneous cardiovascular disease risk prediction federated learning optimization method and system, relates to the technical field of data processing, and the method comprises the following steps: processing original cardiovascular disease data to obtain standardized feature data; based on the standardized feature data, a differentiated training algorithm corresponding to the data structure thereof is used for local calculation to generate updated encrypted local parameters; the updated encrypted local parameters are transmitted to a parameter server in a federated learning framework, and a geometric mean algorithm based on secure multi-party computation is used by the parameter server to securely aggregate and geometrically average the local parameter updates from different sources in an encrypted state to generate a preliminary global parameter set. The application strengthens data privacy protection, improves multi-source heterogeneous data fusion efficiency, optimizes parameter integration to enhance model performance through the federated learning framework.
Need to check novelty before this filing date? Find Prior Art