Federated Learning and Privacy-Preservation in Healthcare
Abstract
Nowadays, due to the increasing demand for Machine Learning (ML) to solve various real-world problems, a substantial amount of sensitive and personal information must be collected and analyzed. This leads to critical data protection challenges. Privacy-enhancing technologies (PETs) are often proposed for data protection, encryption, and Artificial Intelligence (AI) while preserving personal data and ensuring a high level of reliability. The main reason behind data breaches is the ability of deep neural networks to access information. One must realize that, in a standard PETs application, the private data of patients cannot always be fully protected. The implementation of PETs introduces complexities and performance overheads that may limit their scalability and efficiency. When privacy violations occur in patient data, the privacy system must establish safeguards for users’ raw data. We are proposing a hybrid model for protecting patients’ private data. In this model, we utilize algorithms such as Federated Learning (FL) and linear regression. FL enables multiple decentralized devices (such as smartphones or other edge devices) to collectively train a model without the need to share their local data. Linear Regression operates as a prediction method for continuous numerical response variables. The concept behind this model is to ensure private and efficient ML models are implemented in healthcare to secure patient data.
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.