AI-based diuretic self-therapy in heart failure

Heart failure (HF) is the most common reason for hospital admissions of aged patients in developed countries. Heart failure is characterised by episodes of sudden worsening of the symptoms, called decompensation. The majority of hospital admissions belong to patients with decompensation of chronic heart failure. In order to prevent decompensation, telemedicine monitors HF patients for early detection of worsening. However, therapeutic interventions for HF, mostly adjustments in diuretic therapy, are performed by healthcare professionals and self-care is limited. We will develop a decision support system (DSS) based on artificial intelligence (AI) to enable HF patients to adjust their diuretic therapy by themselves. We will combine expert knowledge and knowledge extracted from clinical data using machine learning algorithms to build a model that can produce personalised predictions of the risk of decompensation for a patient given their clinical data, their symptoms, and current therapy. By combining these predictions and preferences elicited by clinicians and patients, our DSS will be able to provide timely personalised advice such as changing the patient’s diuretic therapy or visiting the clinic. This will increase the quality of self-care of patients with HF, prevent unnecessary consultation with healthcare providers and reduce the risk of hospital admissions due to cardiac decompensation, thus reducing costs and healthcare resource use. We intend to implement this eHealth solution as a module that could be integrated into different telemonitoring applications.