2024 – 2029
Funding : France 2030
Limics is a partner in the PostGenAI@Paris cluster, a project led by Sorbonne University, which was selected from among nine research centers in the « IA-Cluster » call for expressions of interest, with a budget of €35 million over 5 years.
https://scai.sorbonne-universite.fr/public/news/view/ca27c448b95463fedf46/7
“Itou” is part of this cluster and concerns the automatic creation of cohorts of similar patients based on medical reports from the AP-HP health Data Warehouse (to date, 120M text reports, 1.2B biological entries). The similarity between patients corresponds to a similarity of symptoms, chronic diseases, biological markers, etc… This project is a continuation of work already carried out in the prototyping phase, with the central aim of staying close to clinical reasoning.
The primary challenge lies in effectively adapting and applying generative models to
real-world data and scenarios. This involves addressing research issues related to data
quality, domain specificity, and the scalability of these models, in order to demonstrate the
real-world utility of generative models in healthcare. This includes addressing challenges
related to deployment, performance, user-friendliness but also ethics.
Maintaining data privacy, sourcing the extracted information, developing mechanisms for
validating the accuracy of extracted information, will be part of the project.
We will evaluate and apply our algorithms to three main classes of use cases: Patient
pre-screening (e.g., for offering genetic screening recommendations or targeted therapeutics), diagnostic support (in numerous clinical scenarios characterised by diagnostic ambiguity), and therapeutic support.
These efforts will ultimately lead to improved healthcare outcomes, enhanced preventive
measures, and more efficient clinical decision-making, advancing the field of precision
medicine.
Contacts: Christel Gérardin, Xavier Tannier