Collaborative Prediction Scheme for Privacy-Preserving Data Based on Vertical Federated Learning

With the proliferation of digital economy, data fusion applications drive innovation across industries. Yet privacy protection poses challenges. In response to this issue, this paper proposes a vertical federated learning scheme based on logistic regression for collaboration between hospitals and schools to predict student depression risk. It utilizes Bind RSA for sample alignment without raw data sharing, Paillier homomorphic encryption for secure data transmission, and RSA for secure exchange of prediction values, ensuring the security and confidentiality of private data for both hospitals and schools. This scheme enhances depression prediction safety and efficiency, addresses privacy concerns, and provides useful guidance for data fusion applications in other fields.