L’Express has placed the defense of rationality at the heart of its DNA. This is why, every week, we highlight the benefits that research bring to society, without ever hesitating to make our contribution to the fight against scientific disinformation. We have chosen, this year again, to extend this commitment by supporting researchers who share these fights thanks to a dedicated event: the Awards for Science and Health Personalities.
The portrait of Sarah Watson could start with a scene from her daily life at the Institut Curie in Paris, where she works. The text, full of praise – it is necessary, to describe the character – would then speak of a “scientist, doctor and at the same time, researcher hurrying in the pale corridors of the hospital, mask on the mouth, round glasses on the nose”. It would be read in action, scribbling medical reports before going to see her patients.
Relive the entire Science and Health Symposium of L’Express 2025 here
Such a primer would be realistic, but somewhat reductive. Describing Sarah Watson as a professional religiously invested in her work is not false, but misses one of the keys to understanding the young woman’s success: deeply “nonconformist” – she says herself – herself – herself – herself – she herself – The scientist never ceases to surprise, to break the codes, making her a pioneer in her field.
At only 39 years old, this cancerologist has developed an artificial intelligence (AI) capable of recognizing the cellular profile of a tumor, and above all, its original organ. A major advance: this information is essential to determine the treatments to be administered against cancer. Until now, in France, around 7,000 patients were without therapeutic solution each year, because doctors could not know in which part of their body the disease was born. Very quickly, many have adopted the tool, in the hope of being able to treat them.
Additional years of life
With its AI, Sarah Watson proves that it is possible to act against “primitive unknown sarcomas”, cancers of her specialty. No one dared to think about it before. “We are talking about years of additional lives for certain patients won thanks to our technology. It is a lot, for this type of disease.” The young specialist also shows that conjugating research and care is possible. His teachers were septic. She knew how to convince them, by dint of work. The Curie Institute gave it a waiver. “I had to keep” real “in my job, build a relationship with the patient,” says this great empathetic.
His artificial intelligence project was born in 2018, a meeting. One of his students, Julien Vibert, very interested in new technologies, tells him about the advances of AI, and his potential health applications. The scientist had never touched this kind of algorithms; Chat GPT did not exist. The young man proposes to systematize and automate his advanced research on the molecular composition of tumors. With a lot of flair, and a certain sense of opening and interdisciplinarity, Sarah Watson accepts.
The duo is falling out for hours, piled up in the scientist’s windowless office. The room measures in everything and for a whole nine square meters, it is covered with posters of painters, Van Gogh, Picasso, Nicolas de Staël, one of the New Yorker and postcards. Sarah Watson comes out of old samples from the Institute’s reserves, subjects them to the machine. Faced with his first evidence of concept, published in 2021 in the scientific journal Journal of Molecular Diagnosticsthe researcher barely achieves the magnitude of their discovery: “We thought that, anyway, we held something interesting,” she recalls. We imagine him celebrating his advance by taking a puff of the vapoteer who hangs out near her computer.
When they are asked about her greatest successes, it is a whole different scene that she evokes. June 2024, at Black Lab, a Lille performance hall. In the middle of a rock’n’roll concert, the researcher improvises Speakerine on stage, for a charity. Clear voice, white t-shirt and proud look, she is applauded by several hundred people. One of the most important moments in her life, she said. That evening, she received a check for 80,000 euros, collected by one of her patients, Charles Delobel.
Background harvest and rock’n’nroll
Their collaboration is worthy of a bad telefilm. She announces the worst: an unknown primitive sarcoma, advanced stage, very little chance of remission. He does only in his head, refuses his spell and plates everything for a last nose. He embarks on the organization of charitable concerts to support “this doctor who does not give the feeling of being one, hyperaccessible, hyperhuman, cash and simple”. Today he is the first financier of his laboratory.
When he told her about the idea, the researcher accepted without hesitation, a guarantee of her taste for adventure, whether scientific or humanitarian, and a certain sense of party: “Sarah Watson is brilliant in every way, but also cultivates a certain originality, a certain freedom. It is not the last to prolong the evening until no hour. Gouill, director of the hospital ensemble at the Institut Curie.
It is this same Charles Delobel that allows Sarah Watson to go from the simple computer prototype to the deployable tool. At the time, Sarah Watson was only at the start of her work. Charles Delobel worked in the Orange telecommunications group. He knows developers in artificial intelligence, knows where to knock on access to the necessary computing power. The researcher feels that it must be taken seriously, accepts her help. The company has since dispatched to it two of its full -time operators to lead the machine to recognize convincing associations Among millions of genetic and molecular characteristics.
“I do not know how she does to get involved as much, stay so close to her patients, hold despite the dramas that are played. It is an enigma,” says the patient, when we talk to her about the commitment of the scientist. Available until very late, she does not fail to send little words to her patients, help them in the test. The job is hard: in the end, very few survive – these types of cancer have a prognosis of the order of a few months at most. “That’s why I do research next to it,” she explains. To feel useful, move forward, and change these bad results in the long term. “
A very special AI
Usually, scientists lead their AI on an exercise and correct them if they are mistaken. Sarah Watson does things upside down. “We first developed an algorithm capable of saying where tumors came from easy to identify and for which we had the answer. Then we submitted our mysterious cancers. The machine released a response, whose reliability we did not know. But that allowed us to choose a treatment and, when we attributed it to the patient, it worked,” said the researcher. ” The specialists reproduce the experience, once, then two, then on fifty patients.
This is perhaps the craziest in this technique: even today Sarah Watson does not know if the result given by the machine is the right one, if the cancers it tests come from the lungs, the pancreas or the kidneys, nor why the algorithm leads to this result. At the head of a steering committee on this technique, the young woman is now working on a new version of her algorithm. It tries to increase its reliability – 20 % of the results are still incoherent. And she already has a new idea: “We could do retro-engineering, that is to say study our algorithm instead of cancers. If we manage to understand how it gives these results, perhaps we can better understand the disease.”
An article in our special file “Personalities of L’Expresss the 2025 prices for science and health”, published on March 13.