학술논문

An Effective FL-CNN Based Data Securing Model for Heart Disease Prediction
Document Type
Conference
Source
2023 6th International Conference on Contemporary Computing and Informatics (IC3I) Contemporary Computing and Informatics (IC3I), 2023 6th International Conference on. 6:1862-1866 Sep, 2023
Subject
Communication, Networking and Broadcast Technologies
Components, Circuits, Devices and Systems
Computing and Processing
Fields, Waves and Electromagnetics
General Topics for Engineers
Photonics and Electrooptics
Power, Energy and Industry Applications
Robotics and Control Systems
Signal Processing and Analysis
Heart
Data privacy
Federated learning
Computational modeling
Cardiac disease
Data models
Convolutional neural networks
Convolutional Neural Network
Federated Learning
UCI Dataset
Cardiovascular disease
Centralized Learning
Language
Abstract
Cardiovascular disease is the leading cause of death worldwide, according to the WHO. Coronary heart disease is most dangerous. 2015 saw 360,000 US heart attack deaths. Effective heart disease treatment prevents global deaths. An updated FL-CNN model improved cardiac disease diagnosis and prognosis for doctors and patients. Hospitals cannot disclose patient data for security and privacy reasons. Thus, centralizing data is hard. Federated Learning can train machine learning and deep learning models using massive volumes of distributed data. On the UCI Cleveland dataset, CNN with Federated Learning achieves 94.99% accuracy, while CNN with centralized learning achieves 97% accuracy.