Module
Artificial intelligence and large language model
Teaching conducted in English
Artificial intelligence (AI) and in particular Large Language Models (LLM) are transforming research practices (e.g., automated information extraction, data cleaning, systematic review support, thematic analysis in qualitative research, assistance in analyzing algorithm).
This course will equip researchers and professionals in the field of clinical research, epidemiology and public health with the conceptual foundations needed to understand how AI systems function, evaluate their reliability, and understand how they could transform research practices.
Participants will gain a solid grounding in key concepts and terminology, as well as the ethical and regulatory considerations surrounding AI. They will also examine the practical challenges associated with real-world data and decision-making constraints. The course introduces a range of AI-based methods and equips participants with practical approaches to identify the most appropriate tools for their specific research objectives.
The course covers four key areas:
- AI key concepts: What AI can and cannot do, model families, model evaluation, bias, uncertainty, and regulatory issues (GDPR, EU AI Act).
- Data & Methods: Characteristics of public-health data, data quality challenges, example of tools and how to choose them.
- Modern LLM Techniques: How LLMs work, prompting strategies, Retrieval-Augmented Generation (RAG), and agents.
- Applied Use Cases: diagnosis, systematic review support, information extraction etc.
No coding or mathematical background is required. The course focuses on conceptual clarity and practical understanding. A basic knowledge of python or of a programming language is helpful, but not required.
General information
About the course
Dates
Prices
Programme
Day 1
Morning session:
- Understanding AI: Concepts and Vocabulary
- Evaluating and Trusting an AI System
Afternoon session:
- Ethics, Regulation & Governance in AI
- Collab and python crash course
Day 2
Morning session:
- Data sources and format
- Data processing methods
Afternoon session:
- Example of tools
- Which methods for which usage
Day 3
Morning session:
- Understanding Large Language Models
- Retrieval-Augmented Generation (RAG) 1/2
Afternoon session:
- Retrieval-Augmented Generation (RAG) 2/2
- Agents
Day 4
Morning session:
- Some Real Use Cases
Afternoon session:
- Hands-on / Question Session
Speakers
Mélissa Duran
Mélissa Duran is a general practitioner and doctoral student in clinical epidemiology. She is driven by a deep passion for primary care, critical thinking and evidence-based medicine. With her medical training and a master’s degree in comparative research on effectiveness, she joined the METHODS team at CRESS-UMR1153 in order to deepen her expertise. She is currently preparing a doctorate in clinical epidemiology (under the supervision of Professor Boutron and Professor Sidorkiewicz). Her work focuses on both the use of AI in general medicine and the use of AI in clinical epidemiology.
Alex Fernandes
Alex Fernandes is a doctoral student at Université Paris Cité and an instructor at the École Normale Supérieure – PSL. His research focuses on the use of statistical learning for causal inference. He first earned a degree in mathematics, then rounded out his education by studying biology and public affairs. He worked on algorithms for the topological analysis of data and the effect of individualized processing, and participated in the development of several tools, particularly in the field of deep learning. His thesis aims to develop a framework for linking randomized controlled trials to the real population using registry data, as required for health technology assessment.
François Petit
François Petit is research director at Inserm. Holder of a doctorate in pure mathematics, with a specialization in algebraic geometry, he initially worked in the fields of algebraic geometry and mathematical physics. His work then evolved into the development of data analysis methods based on algebraic topology techniques. In 2019, he obtained a chair of excellence from Université Paris Cité. His current research focuses on the development of mathematical, statistical and machine learning methods for personalized medicine, with a particular focus on causal inference and topological data analysis applied to complex biomedical data.
Thomas Starck
Thomas Starck is a teaching and research officer. His doctoral dissertation focused on the application of data analysis methods to environmental issues. He then moved into meta-research and joined the METHODS team at CRESS, where he studies the evolution of the certainty of evidence relating to medical interventions and develops interventions aimed at strengthening the primary research ecosystem. His work also focuses on automating evidence synthesis processes. At the same time, he teaches statistics and epidemiology to medical and master’s students.