Skip to main content

Raisaro Group

University of Lausanne

Our group, led by Prof. Jean Louis Raisaro, assistant Professor at the Faculty of Biology and Medicine of the University of Lausanne, develops and translates innovative AI and data science solutions to improve clinical care and enable multicenter research. We work across trustworthy AI methods, including privacy-preserving approaches such as differential privacy and synthetic data generation, clinical Large Language Models (LLM) and agentic systems, as well as multimodal foundation models that integrate clinical data from electronic health records, including clinical text, structured variables, and time-series signals.

We combine reproducible methods research with applied clinical projects, and we develop platforms that accelerate safe, ethical, and regulatory-ready AI deployment. These include CHORUS, a secure processing environment enabling compliant access to CHUV biomedical data and clinical AI development, and an Agentic AI platform supporting human-in-the-loop AI-assisted workflows. Several of our solutions are currently being implemented at CHUV, with measurable impact on patient outcomes, efficiency, and hospital costs. Recent achievements include an AI-enabled Sepsis Learning Health System and Meditron-CHUV, an open-source medical LLM fine-tuned on more than 12,000 clinician preference annotations from over 250 CHUV clinicians.

Our team

Jean Louis Raisaro

Direction de l'innovation et de la recherche clinique, Tenure track assistant professor

Jérémie Despraz

Direction de l'innovation et de la recherche clinique, Principal data scientist

He Xu

Direction de l'innovation et de la recherche clinique, Senior data scientist

Giorgia Carra

Direction de l'innovation et de la recherche clinique, Senior data scientist

Nathalie Casati

Direction de l'innovation et de la recherche clinique, Senior software engineer

Guy Spuhler

Direction de l'innovation et de la recherche clinique, Senior software engineer

Sami Perrin

Direction de l'innovation et de la recherche clinique, Senior software engineer

Jérémie Frei

Direction de l'innovation et de la recherche clinique, Junior Full-Stack Software Engineer

Bogdan Kulynych

Direction de l'innovation et de la recherche clinique, Research scientist

Bernd Illing

Direction de l'innovation et de la recherche clinique, Data scientist

Mohamed Kaabachi

Direction de l'innovation et de la recherche clinique, PhD student

Marzio Cossali

Direction de l'innovation et de la recherche clinique, PhD student

Farah Briki

Direction de l'innovation et de la recherche clinique, Junior data scientist

Aleksandra Novikova

Aleksandra Novikova

Direction de l'innovation et de la recherche clinique, PhD student

Snezana Nektarijevic

Direction de l'innovation et de la recherche clinique, Cheffe de projet

Projects

  • CHORUS: enhancing secure biomedical data access

    CHORUS, a secure processing environment developed at CHUV, enables reliable sharing of biomedical data and collaboration among research teams. By combining advanced encryption with rigorous authentication protocols, CHORUS promotes biomedical innovation while protecting sensitive patient information.

    Find out more

  • MED.USE : Medical large language model for reducing resource overuse in the emergency department

    We develop a decision-support system based on a medical language model, capable of processing commonly available information and providing recommendations to emergency department clinicians, with the goal of reducing low-value care.

    Find out more

  • SYNTHIA: synthetic data generation framework for healthcare applications

    This project aims to promote the responsible use of synthetic data in health research and innovation. It addresses challenges related to data confidentiality and complexity through innovative techniques for generating multimodal and longitudinal datasets. The open platform SYNTHIA will provide validated synthetic data and tools to advance research without compromising patient confidentiality.

    Find out more

  • MOOVE-CHUV: improving medical AI with real‑world clinical insights

    We have developed Meditron-CHUV, an open-source medical Large Language Model, fine-tuned on more than 12,000 clinician preference annotations from over 250 CHUV clinicians. We introduce a clinician-centric framework that positions human expert participation as a central pillar of LLM development and evaluation, establishing a feedback-loop system where expert evaluations simultaneously assess model performance and generate high-quality alignment data for model optimization.

    Find out more

Selected publications

  • (opens a new window)