AI tools for automating personalised and participatory evidence-based treatment recommender systems

ENG- In recent years, evidence-based medicine has transformed how healthcare professionals make decisions, by systematically evaluating treatments according to scientific research. However, traditional approaches still face important limitations: they are not fully automated, they rarely consider the preferences of patients and clinicians, and they often lack personalized recommendations. This thesis addresses these challenges by developing new methodologies that enhance automation, participation, and personalization in therapeutic recommendations. The research focuses on attention deficit hyperactivity disorder (ADHD), a complex condition that requires individualized treatment strategies. First, it introduces APPRAISE-RS, a system capable of automatically analyzing medical literature and generating treatment recommendations following the international GRADE standard. Unlike conventional methods, APPRAISE-RS integrates patients’ clinical data and incorporates their preferences regarding drug effects, producing more transparent and personalized recommendations. Second, the work applies multi-criteria decision analysis (MCDA) techniques, which combine clinical expertise with patients’ priorities. This ensures that treatment choices reflect individual needs and values, fostering a shared decision-making process between patients and healthcare professionals. Finally, the thesis proposes an innovative drug grouping methodology using machine learning to cluster medications with similar properties and effects. This strengthens the synthesis of scientific evidence and leads to more robust and reliable recommendations. The results show that it is possible to move towards a more automated, participatory, and personalized evidence-based medicine that improves the quality of treatment recommendations. While ADHD served as the case study, the proposed methodologies can be extended to many other medical conditions, paving the way for a future in which patients and clinicians decide together on the most appropriate treatments, supported by stronger and more tailored scientific evidence