Federated Attention-Based Explainable Deep Learning Framework for Real-Time Cardiovascular Disease Prediction Using Wearable IoMT Data

Karimunnisa shaik -, P. Sridhar, Praveen -

A Federated Attention-Based Explainable Deep Learning Framework for Real-Time Cardiovascular Disease Prediction using Wearable IoMT Data is an innovative healthcare model based on a distributed system of wearable sensors to predict cardiovascular diseases mainly focusing on patients' privacy and data security. The clinical trustworthiness is, however, lowered due to the difference in the type of data in an IoMT device and the absence of a comment to explain the predictions of deep learning. The proposed framework proposes the adoption of federated learning, the incorporation of an attention-based CNN-LSTM architecture, and XAI methods such as SHAP and LIME to improve the forecasting accuracy of the diseases while ensuring security and transparency. The proposed system helps to attain greater prediction accuracy, real-time monitoring, secure patient sensitive data, and to enhance model interpretability and clinical decision making speed and reliability.