On April 1st, 2025, the developed Medical Decision Support System (MDSS) was clinically validated at the University Hospital of Patras, marking a significant milestone in the project.
Under the scientific coordination of Professor Elpiniki Papageorgiou (Principal Investigator), the validation was supported by members of the EMERALD team, including Associate Professor Nikolaos Papandrianos, Professor of Nuclear Medicine Dimitrios Apostolopoulos, Postdoctoral Researcher Dr. Ioannis Apostolopoulos, and Associate Professor Theodosis Theodosiou. The evaluation involved the use of anonymized clinical records referring to Coronary Artery Disease and Non-small Cell Lung Cancer for evaluating the functionalities and diagnostic performance of the MDSS. Real diagnostic workflows were simulated, closely mirroring the conditions faced by healthcare professionals in practice.
Clinicians assessed the system’s performance based on diagnostic accuracy, interpretability, and usability. The MDSS’s explainability tools were especially valued for providing transparent, understandable justifications for each prediction. Additional features, such as the natural language generation module, further enhanced the system’s usability and integration into clinical routines. This collaborative effort between AI researchers and healthcare professionals ensured that the system’s capabilities were aligned with practical clinical expectations and needs. The pilot deployment validated the MDSS as a clinically reliable and ethically grounded decision-support tool. It demonstrated high predictive precision and received strong endorsement from medical experts, who emphasized its potential to streamline diagnostic workflows and reduce diagnostic uncertainty.
The clinical team’s feedback provided valuable recommendations for future refinements, aimed at broadening the system’s applicability and facilitating deeper integration into diverse healthcare environments. These findings reinforce EMERALD’s mission to deliver impactful, interpretable, and clinically relevant AI solutions for medical decision support.
