Security and Privacy in Federated Learning–Based Internet of Medical Things
Abstract
The current global pandemic has increased the development and usage of mobile, wearable and other digital technologies for collecting the health-related data of a person. The digital data thus collected could be used to identify useful information, using machine learning techniques, which would help to assist the doctors in early prediction and diagnosis of diseases. This has indirectly led to lack of privacy and personalization of patient data. Hence, federated learning (FL) could be used to build a healthcare framework for organizations that mitigates the risk of compromising users’ privacy and security. It trains a model across multiple edge devices in the medical data centers holding local data samples, without exchanging them. This chapter discusses a FL-based healthcare system that is secure and protects user privacy. The framework is proposed using the Paillier cryptosystem. It uses additive homomorphism for the aggregation and training of data in the global data centers. Initially, the model collects data specific to each organization and trains the model locally. On a periodic basis, the data available in each data center is encrypted and sent to the global data center. At the global data center, the data from the wearable resources and the medical data centers are aggregated using unique identifiers. A public key cryptography with additive homomorphism is applied to ensure secure data aggregation and training in the global data center. A trusted party is involved…
Read the paper
Related studies
- Active Learning during Federated Learning for Object Detection
- Fairness-aware federated learning
- Decentralized Federated Learning
- Building a Personally Identifiable Information Recognizer in a Privacy Preserved Manner Using Automated Annotation and Federated Learning
- Preservation of Patient Level Privacy: Federated Classification and Calibration Models
- Efficient explainable federated learning
Go deeper in Evidence Hub
Evidence Hub adds a study data sheet (design, sample size, evidence level, key numbers), an AI summary and a chat grounded in the paper's text.