Towards a Knowledge-Based Decision Support System for the Management of Type 2 Diabetic Patients

The rise of Healthcare 5.0 paradigm calls for personalization of care and management of patients' conditions. Though promising, data-driven techniques may raise some concerns as they are perceived as scarcely transparent and reliable by clinical personnel. With the emergence of Explainable Artificial Intelligence (AI), these limitations could significantly be overcome. In this regard, the exploitation of domain knowledge (properly formalized) can support explainable AI and foster the delivery of Decision Support Systems (DSS) for tailored treatment of many diseases. This work aims to present a knowledge-based DSS for managing patients with Type 2-Diabetes Mellitus (T2D), a non communicable disease that can take advantage of tailored medical nutrition therapies, taking into account patient's specific health condition and comorbidities. The DSS leverages ontological representation of domain knowledge to automatically classify the patients' phenotype and identify the potential comorbidities, then, it exploits a set of rules to provide tailored nutrition recommendations that can be adopted by general practice doctors and family clinicians to provide tailored dietary plans. In this way, the proposed DSS can support physicians and dieticians (who may lack specialized training in T2D management) in the management of diabetic patients through personalized medical nutrition therapies.