This is a booming field, the possibilities of which the general public and part of the medical profession struggle to imagine, as this world of algorithms is so foreign to them. However, in Limoges, researchers and clinicians have begun to integrate these tools into their work and practices.
All medical specialties will one day be impacted by artificial intelligence (AI). A revolution in which researchers and health practitioners in Limoges are interested. On October 17, 2024, the inaugural back-to-school conference of the Faculty of Medicine and Pharmacy, intended to launch the academic year, was devoted to AI.
Jean-Baptiste Woillard, professor of pharmacology, gave the presentation. He looked into the subject in 2015, not without difficulty. “At first, I told myself I wouldn’t make it. There were many new concepts and the means to democratize these methods had not yet been found. »
Scientific mobility in Toulouse and Cambridge has allowed him to progress, and the release of new, more powerful mathematical algorithms since 2020 has allowed him to consider solving certain problems in his work as a researcher.
How does AI help you in your research?
“In pharmacology, I am working on the development of an application that would allow individualizing the dose of anti-rejection drugs (in the case of transplantation) and anti-infectives (in a context of antibiotic resistance). The idea is to give the right dose to the right patient, at the start and during treatment by integrating dynamic patient data in order to refine the therapeutic strategy. This is a complex, multifactorial problem that involves predicting drug concentrations.
For two years, as part of the Priority Research Program and Equipment (PEPR) in digital health*, I have been coordinating a scientific consortium with various partners on “pharmacological digital twins”, for which we have obtained funding of 1.8 million euros. Concretely, by grouping and analyzing real clinical data from anonymized patients, AI allows us to generate “twin” data which integrate different types of biological models (molecular, tissue, genomic, blood) and which take into account the environment and the patient’s habits…”
With this synthetic data, the goal
is to predict the effectiveness of drugs,
to improve the benefit/risk balance
and limit side effects.
Receive our eco newsletter by email and find the latest news from economic players in your region.
What place will artificial intelligence have in medical practices?
“It will help address major challenges in medicine. There is no shortage of examples. AI will be able to make individual prediction much more effective than the statistical approach. We can say that the average risk of having a myocardial infarction when you are a smoker increases by a certain percentage, but with AI, based on a series of patient characteristics, it is the individual risk that will be evaluated.
In cancer research, AI will save precious time by predicting with the same precision the results of a clinical trial well before the end of said trial…
Being able to integrate different sources and complex data into an algorithm, which do not evolve in the same way over time and which integrate the specificities of each individual, is a considerable advance. This generates a lot of hope, since it will make it possible to improve the diagnosis of the disease, to personalize the treatment…”.
Are doctors already trained in these new approaches?
“It is important to develop training now. Two years ago, we created a “AI in health” teaching unit at the Limoges medical faculty for 3?, 4? and 5? years, in the initial training course.
In November 2024, with Doctor Marc Labriffe, we launched a university diploma “Artificial Intelligence in Pharmacology”, as part of continuing education.
Workshops are also planned for 2025 in pharmacometrics, this discipline which studies the evolution of the life of the drug once administered. »
Artificial intelligence also raises fears. Will she replace the doctor?
“Of course not, it will be more of a man/machine collaboration. In people’s minds, AI self-learns and grows on its own, but this is rarely the case (except for reinforcement learning approaches like chess).
Humans power it and AI is a decision-making tool, which can make mistakes, which has biases and limits.
The man – and therefore the doctor – will have the last word and take responsibility. But the profession will evolve. Tasks with low added value can be automated, which will free up time for more complex missions.
In ophthalmology, for example, as part of mass screening for diabetic retinopathy, the machine will be able to select patients at risk from thousands of images, and the specialist will only have to concentrate on these people there. It’s an opportunity.
We can talk about the notion of an augmented man. And if machines do not replace humans, doctors who use AI will replace those who do not. »
This program is led by Inserm and Inria (National Institute for Research in Digital Sciences and Technologies). [Therapixel]MammoScreen
artificial intelligence that detects breast cancer
Prognosticating autism from birth using artificial intelligence: a study carried out in Limoges
