This French start-up wants to use AI to develop new medicines more quickly

The French startup partnered with Amazon Web Services (AWS) to develop a generative AI application to assist pharmaceutical companies ©Canva

It takes a decade on average before a new medication arrives on the market.

Now, the French start-up Yseop is trying to use generative artificial intelligence (GenAI) to speed up the process.

Clinical trials often generate a huge amount of data, and it’s the job of medical writers to generate clinical trial protocols and final reports.

"They have to report how the clinical trial went, who the participants are, the details of the participants and they have to give details [about] the manufacturing, the stability of the manufacturing, the quality [of the trial and drug]," Emmanuel Walckenaer, Yseop’s CEO told Euronews Heath, adding that around 30 report types exist ranging from 10 to 1000 pages.

Accuracy is essential as the documents are sent to health authorities to secure approval for the trials and marketing authorisations.

"What we write can't be approximate, it has to be perfect," Walckenaer said.

Medical writers could use GenAI to speed up the writing process and bring the drugs to the market faster, he claims, adding that Yseop is working with several AI models from companies such as Mistral, OpenAI, and Anthropic.

Yseop’s clients, such as pharmaceutical giants Eli Lilly and Company and Sanofi, have said that it helps efficiency as the medical writer can work more quickly, according to Walckenaer.

He added that there is a shortage of medical writers, so the start-up wants to help them speed up their more tedious tasks such as reading statistical tables.

How does Yseop guarantee accuracy?

Yseop uses both text-to-text GenAI as well as data-to-text when it comes to managing statistics.

"We can't afford to have hallucinations or omissions,” Walckenaer said, explaining that the company implanted high-level quality checks to prevent errors.

"The volume of data to be processed almost triples per year," he said.

The company is partnered with Amazon Web Services (AWS) to handle some of the technical challenges such as data storing and localisation.

A strict quality process was developed internally by several people with criteria such as language quality or accuracy with a minimum threshold to prevent any mistakes.

In addition to that process, the medical writer also double-checks the output and the pharmaceutical company has its additional verification layer.

"Today, we are working with five of the 15 largest pharmaceutical companies in the world," Walckenaer said, adding that one of the company’s goals is to target middle-size clients.

AI is slowly making its way into medicine

Yseop is in the vanguard of a growing momentum to implement AI in the field of medicine.

American hardware company IBM used its computer system Watson towards medicine. Last year, its health data and analytics assets were acquired and became a standalone company: Merative.

At the beginning of the year, Israel-based AI company QuantHealth, founded in 2020, raised $17 million (€15.7 million) in a funding round to propel its drive for AI-driven clinical trials.

Big pharma companies such as Amgen, Bayer, and Novartis are also betting on AI speed up clinical trials by scanning huge amounts of data about potential trial patients, such as public health records, prescription data, medical insurance claims, and their internal data, according to Reuters.

In some cases, this time to gather a cohort could be cut by 50 per cent, the companies claimed.

"I don't think it's pervasive yet," Jeffrey Morgan, managing director at the consulting company Deloitte, which advises the life sciences industry, told the news agency.

"But I think we're past the experimentation stage".

© Euronews