EMERALD takes a holistic approach to patient-specific predictive modeling and eXplainable AI (XAI)-based Medical Decision Support Systems (MDSS) development by integrating knowledge from new research, clinical tests and Health electronic Records. To achieve this, EMERALD uses advanced analytic techniques (Data Mining, Deep Learning (DL) and Advanced Fuzzy Models), enabling the analysis simplification of voluminous patient data and thus allowing the development of personalized predictive XAI-based MDSS. Model-driven MDSSs with the focus on explainable analytics incorporating the dynamic features of Fuzzy Cognitive Maps (FCM), will play a major role in EMERALD. EMERALD will further introduce the new concept of XAI-based DeepFCMs as the innovative and structural component of an XAI-MDSS. DeepFCMs will:
(i) fuse a multitude of medical data spanning from text to
images,
(ii) be executed as typical deep neural networks and
(iii) be presented visually as FCMs providing visual explanations and reasoning.
Finally, EMERALD aims to create a medical science ecosystem that will provide physicians with effective explanations, focusing on data interpretability and accuracy. Two common complex
diseases in Nuclear Medicine, the Coronary Artery Disease and Non-Small Cell Lung Cancer (NSCLC) have been selected as use cases to show the capabilities of the new XAI-based MDSS. EMERALD is expected to offer the medical care sector a solution of personalized XAI-MDSSs capable of:
(a) identifying the indicators that predict the likelihood of future events (e.g. ischemic episodes, heart attacks, mortality),
(b) predicting and classifying various heart failure incidents and NSCLC, and
(c) adopting a novel holistic approach to patient-centered predictive model that can create multidimensional impacts on Greek and European economy, society and healthcare industry.