French start-up Yseop is leveraging generative artificial intelligence (GenAI) to expedite the lengthy drug development process. Partnering with Amazon Web Services (AWS), Yseop aims to assist pharmaceutical companies in navigating the complex stages of clinical trials and drug approval more efficiently.
Typically, bringing a new medication to market can take about ten years. Clinical trials generate extensive data that medical writers must compile into detailed reports for health authorities to secure trial and marketing authorisations. Emmanuel Walckenaer, Yseop’s CEO, explains that the accuracy of these documents is critical, as they must meet stringent regulatory standards.
Yseop’s GenAI tools are designed to help medical writers speed up the creation of clinical trial protocols and final reports. This is particularly beneficial given the shortage of medical writers and the increasing volume of data to be processed, which triples annually. Pharmaceutical giants like Eli Lilly and Sanofi have reported increased efficiency, allowing medical writers to focus on more complex tasks rather than data-heavy, repetitive ones.
To guarantee the precision of its AI-generated documents, Yseop employs both text-to-text and data-to-text GenAI models. The company has implemented rigorous quality checks to prevent errors such as hallucinations or omissions. These checks are supplemented by manual reviews from medical writers and additional verification by the pharmaceutical companies themselves.
Yseop’s collaboration with AWS addresses technical challenges related to data storage and localisation. The start-up is currently working with five of the world’s 15 largest pharmaceutical companies and aims to expand its client base to include mid-sized firms.
Yseop is at the forefront of integrating AI into the medical field, joining other companies like IBM’s Watson and Israel-based QuantHealth in the drive for AI-driven clinical trials. Major pharmaceutical companies such as Amgen, Bayer, and Novartis are also investing in AI to streamline the patient recruitment process, potentially cutting cohort gathering times by 50%.
