Área: Engenharia Informática e de Computadores

Despacho de nomeação de Júri)

Título da Tese: Towards AI-enhanced Cancer Multidisciplinary Team Meetings

Local da Prova: Anfiteatro PA-3 (Piso -1 do Pavilhão de Matemática) do IST

Data: 22/04/2025

Hora: 09h30
Abstract: Multidisciplinary Team (MDT) based decisions have become the standard of care in oncologic disease management. MDTs operate through a structured pipeline beginning with diagnosis and followed by deliberation in MDT Meetings (MDTMs). These meetings serve as forums for evidence-based discussions, leveraging data presentation and analysis to reach consensus on treatment plans. The concept of collective intelligence, which refers to the emergent knowledge from group deliberation, is central to the efficacy of MDTs. However, MDTs face challenges such as handling unstructured data, limited participation, and technological issues. Artificial Intelligence's (AI) ability to collect and analyze data parallels the collective intelligence process, making it a promising tool for MDTs. In particular, given the reliance on free-text data in clinical settings, LLMs could streamline MDT tasks, improving the quality and efficiency of the MDT's deliberation process. However, the use of LLMs in this context remains largely unexplored. The research employs a mixed-methods approach, including observations, interviews, and specula- tive methods, to understand and develop AI tools tailored to MDT needs. A conversational LLM prototype was designed and evaluated for its impact on MDT deliberation quality. This work contributes in three key areas: understanding current MDT practices and challenges, identifying opportunities for LLM integration, and evaluating the impact of LLM-supported tools on MDT deliberations. The findings provide insights into the potential of AI to transform cancer care MDTs, paving the way for future research and development in this field.