Hybrid Al Framework for Remote Patient Heart Failure Risk Prediction

Heart failure is a lethal disease with a high risk of death. Once it is diagnosed, it cannot be cured definitively. Continuous attention needs to be given to such patients in both hospital and home environment. Due to the lack of measurement and risk analysis at home, the chance of early intervention is dramatically reduced. This research aims to offer a hybrid AI-based approach for patients to monitor and assess their daily risk at home. Highly personalized AI models are trained with continuous data stream from wearable devices and demographic information. The personalized risk assessment is based on individual’s benchmark of the daily routine and bio pattern. Also, the rule-based algorithms are developed based on clinical well-established thresholds. This hybrid approach yields 87.5% accuracy. Our approach can provide a cost-efficient method for at-home patients’ heart failure prediction.