The DDHT school on MEDical Informatics for AI translation from Bench to Bedside (MEDAI) is organized by the University of Pavia (Italy), together with the Stockolm University (Sweden), Brunel University (UK), University of Ljubljana (Slovenia), Otto-von-Guericke-University Magdeburg (Germany), University of Porto (Portugal)

Objectives

The DDHT MEDAI school has the goal of training future leaders of AI in medicine about the dynamic nature of this discipline and the importance of the human-inthe-loop and human-in-command approaches. We envision a future where AI enhances the capabilities of medical specialists, rather than limiting them. To reach this goal, we will organize a PhD school that will boost expertise for the application of AI in medicine, through an active and constructivism learning approach, a training model that asks students to engage in their learning by thinking, discussing, investigating, and creating. Despite the immense potential of AI in healthcare, particularly in prognostic forecasting, significant challenges hinder its practical implementation in real-world clinical settings. This proposal is dedicated to addressing these challenges systematically. Within the school, we will provide the students with relevant knowledge to develop the necessary skills to identify and overcome the barriers preventing the efficient translation of AI prognostic models into clinical practice. Specific emphasis will be placed on how AI models based on temporal dimension and multimodal data integration can tackle these challenges.

We aim to provide PhD students with practical tools for building informatics approaches able to support clinical decisions empowered by cutting edge artificial intelligence (AI) approaches, including Temporal Representation Learning and Multimodal Data Fusion. PhD students will learn how to shape their research for streamlined translation of AI models from Bench to Bedside. We aim at teaching PhD students on how to leverage collaborative environment leading to impactful advancements in AI-driven healthcare. Our initiative will host PhD students from diverse backgrounds in computer science, engineering, and medicine, to tackle the problem from different perspectives and within multidisciplinary settings.

Credits: 2 ECTS

Format and Topics

The summer school will take place over 3 days (Mon - Wed) and be completely taught in English. 

The high-level program for the five days consists in:

Day 1. Presentation from PhD students regarding their PhD project (10 minutes presentation + 10 minutes of discussion with mentors).  Keynote "How to survive a big project coordination: the Capable experience" by Silvana Quaglini and Lucia Sacchi (Pavia). Presentations schedule: docs.google.com/document/d/1AK6tPZZIinHlsxR3eRiofO2iF10QYh4R4DkzWvAQcSA/edit?usp=share_link 

Day 2. Departure from Pavia to Brallo in early morning (transportation guaranteed). Journal club leaded by John H. Holmes.  Working groups in relevant topics (Brallo di Pregola)

Day 3. Students will write a draft of a blueprint on their topic of interest, and prepare a presentation. Time off for biking, hiking, yoga.. (Brallo di Pregola)

Day 4. Closing of the school and returning to Pavia in the morning

Working groups and relevant topics

During the second day, mentors and students will discuss about relevant issues in AI in medicine and define working groups. Each group will be coordinated by a mentor. Mentors will be supporting students through tout group discussions towards the identification of relevant issues and their possible solutions. Throughout the program, students will engage in group discussions, collaborative projects, related to their respective topics. The goal is to enhance their understanding, skills, and practical application of AI in healthcare. 

The relevant focus will be placed on the following topics:

·   Reliability Approaches for AI models (Coordinators: Giovanna Nicora, Ioanna Miliou)

·   Temporal Representation Learning (Coordinator: Allan Tucker, Lucia Sacchi)

·   Multimodal Data Fusion (Coordinator: Panagiotis Papapetrou, Daniele Pala)

·   Biases in Healthcare Data (Coordinator: Myra Spiliopoulou, Arianna Dagliati)

·   Decision Support in Disease Progression (Coordinator: Pedro Pereira Rodrigues,Enea Parimbelli)

·   Summarizing and Visualizing Temporal Data (Coordinator: Blaz Zupan, Pietro Bosoni)

During day 3, the working groups of students will write a draft of a blueprint and prepare a presentation for mentors