This three-day workshop, taking place from the 24th to the 26th of June 2026, will bring together researchers working machine learning and related fields to exchange knowledge and explore advances in Artificial Intelligence for Medicine.
The event is will primarily be open to 4EU+ PhD students in medicine, biology, and computer science who wish to deepen their understanding of AI techniques and their applications in healthcare.
This event is jointly organised by the Sorbonne Cluster for artificial intelligence (SCAI) and Mannheim Institute for Intelligent Systems in Medicine (MIISM) and is financially supported by the international organisation “Franco-German University”.
The aim of the three-day workshop offers young researchers a comprehensive, immersive learning experience. It kicks off on the first afternoon, following participants' midday arrival. The second day is intensive, alternating between expert lectures and hands-on sessions. The third day is dedicated to morning presentations before participants depart.
The multicultural environment encourages the comparison of scientific approaches and a deeper understanding of French and German research contexts. This dynamic is supported by the close collaboration between Sorbonne University and Heidelberg University in the fields of AI and health. The intercultural exchange, covering both scientific content and research practices, provides participants with significant added value for their training.
The program will feature a diverse range of formats, including plenary lectures to introduce the latest advancements in the field, concise presentations and poster sessions showcasing ongoing research, and practical workshops focusing on recent developments. Additionally, roundtable discussions will offer a platform for in-depth dialogue and idea exchange.
The audience will comprise a diverse group of PhD students from the fields of medicine, biology, and computer science. The event has 40 seats available, with 10 reserved for participants from 4EU+ alliance partners and the remaining 30 allocated to students from Heidelberg University and Sorbonne University.
The presentations will be published in a conference volume. Additionally, we'll disseminate project results through articles on the Sorbonne University and SCAI websites, notably via the 4EU+ Alliance, which has significant outreach potential, especially in Europe.
APPLICATION
deadline: March 15th, 2026
Doctoral candidates from 4EU+ Alliance who have a keen interest in artificial intelligence applied to the medical domain are invited to apply for this event. Interested applicants should complete the form by March 15th, 2026.
It's important to note that a minimum level of proficiency in programming skills and English language is required to effectively participate in the workshops activities.
Selected candidates will be asked to give a short presentation of their research activities.
Program
| Day 1 - June 24th, 2026 |
|
12:30 - Reception and Lunch 13:30 - Welcome Presentation: Jurgen Hesser 14:00 - Participant presentations and postersbreaking 15:30 - Coffee break 16:00 - Presentation: Erwan Scornet " Risk ratio, odds ratio, risk difference... Which causal measure is easier to generalize?" There are many measures to report so-called treatment or causal effects: absolute difference, ratio, odds ratio, number needed to treat, and so on. The choice of a measure, e.g. absolute versus relative, is often debated because it leads to different impressions of the benefit or risk of a treatment. Besides, different causal measures may lead to various treatment effect heterogeneity: some input variables may have an influence on some causal measures and no effect at all on others. In addition some measures – but not all – have appealing properties such as collapsibility, matching the intuition of a population summary. In this talk, I will clarify the notions of collapsibility and treatment effect heterogeneity, unifying existing definitions. I will show that only the risk difference has its CATE and ATE (average treatment effect) disentangled from the baseline, regardless of the outcome type (continuous or binary). As we are interested in the generalization of causal measures to target populations that differs from that of randomized controlled trials, we show that different sets of covariates are needed to generalize an effect to a target population depending on (i) the causal measure of interest, and (ii) the identification method chosen, that is generalizing either conditional outcome or local effects. 16:30 - Hands-On-Session 1 18:00 - Dinner 19:00 - Social activity: TBA |
| Day 2 - June 25th, 2026 |
|
9:00 - Presentation: Verena Schneider-Lindner "AI for fighting Sepsis in the ICU: Challenges and Opportunities" Sepsis is the most common cause of death in the non-cardiac intensive care unit (ICU). Artificial intelligence and particularly machine learning (ML) have been applied for sepsis research focusing on prediction of incidence and mortality as well as early diagnosis. Few ML-based models, however, are yet clinically applied on a large scale and their superiority compared to common clinical scores is unclear. I will review existing models and algorithms and present results from our own ML-based studies on sepsis. The focus will be on two main issues impacting the validity and accuracy of sepsis models and thus their clinical utility that are also pertinent to other use cases. The first is the importance of a valid outcome definition and identification for model training, whereas the second is the impact of missing data on model performance. Finally, a much more general potential role of medical AI for supporting sepsis care is proposed. 9:30 - Round Table Discussion 10:30 - Coffee break 11:00 - Hands-On-Session 2 12:30 - Lunch break 13:30 - Presentation: Clement Rambour 14:00 - Participant presentations and posters 15:30 - Coffee break 16:00 - Presentation: Cleo Weis "On Black Swans and Red Herring in Digital Pathology" Diagnostic pathology, whether conventional or digital, faces two critical challenges: Black Swans, rare but high-impact diagnoses (e.g., overlooked Hodgkin cells), and Red Herrings, deceptive mimics (e.g., reactive immunoblasts masquerading as malignancy). These pitfalls risk misdiagnosis with severe clinical consequences. This talk presents a computational pathology framework to systematically address them by distinguishing two key analytical settings: In the police line-up scenario, standardized instances (e.g., glomeruli or lymph nodes) are directly compared using supervised or few-shot learning, enabling classification of well-defined entities. Far more complex is the needle-in-a-haystack setting, where informative regions (e.g., tumor budding in colorectal carcinoma) must first be localized within vast whole-slide images (WSIs) before rare or ambiguous features can be assessed. 16:30 - Hands-On-Session 3 18:00 - Dinner 19:00 - Social activity: TBA |
| Day 3 - June 26th, 2026 |
|
9:00 - Presentation: Isabelle Bloch "Hybrid and Explainable Artificial Intelligence for Medical Image Understanding" This presentation will focus on hybrid and explainable AI, more specifically in the domain of spatial reasoning and medical image understanding. Image understanding benefits from the modeling of knowledge about both the scene observed and the objects it contains as well as their relationships. We show in this context the contribution of hybrid artificial intelligence, combining different types of formalisms and methods, and combining knowledge with data. 9:30 - Hands-On-Session 4 10:30 - Coffee break 11:00 - Hands-On-Session 4 12:30 - Farewell & Lunch |