Hybrid Explainable Artificial Intelligence and Biomedical Signal Processing for Real-Time Prediction of Cardiovascular Disorders Using Wearable Internet of Medical Things Devices

Abstract

Background: Cardiovascular disorders are one of the principal causes of untimely morbidity, and therefore, it is necessary to monitor continuously and in real-time through wearable Internet of Medical Things (IoMT) devices explained by artificial intelligence (XAI). Purpose: The proposed study will present a hybrid explainable model of AI with biomedical signal processing to achieve precise real-time forecasts of cardiovascular diseases based on wearable ECG and PPG signals. Methods: The framework combines signal preprocessing, feature extraction, CNNBiLSTM hybrid model, an attention mechanism, and SHAP explainability. The cardiovascular datasets publicly available were used to assess performance using standard classification measures. Findings: A 98.4% accuracy, 97.9% precision, 98.1% recall, 98.0% F1-score, and 99.2% ROC-AUC were obtained with the proposed framework, which takes only 18.6 ms to infer. Conclusion: The suggested explainable AI-based wearable IoMT system has superior predictive accuracy, real-time, and increased clinical transparency, which makes it an attractive remedy to early cardiovascular disorders detection and continuous remote medicine supervision.