AWS Pioneers Project

Clarisse dos Santos, now Director of Product Management at Yseop, recalls being overwhelmed by paperwork during her post-doctorate in clinical and neuro-imaging data analysis. Yseop’s solution is addressing this very challenge by transforming data into precise regulatory documents.

“When I was writing, I was doing a lot of copy and paste and taking information from different sources. And I kept thinking, I wish I could just have a button to take all these things and put them in the right place.”

Yseop provides this ‘magic button’, converting structured and unstructured data into precise drafts of critical regulatory documents, including CSRs, patient narratives—cutting the process from months to weeks.

Providing the backbone for this is Amazon Bedrock and its support of Anthropic’s Large Language Model Claude 3.5.

“We found that it was the best of all the available LLMs for us when it came to analysis of quantitative data”, says dos Santos. "Medical writers can adapt the AI model to their specific writing style. They are particularly pleased with the model's accuracy, especially regarding statistical data, because one of the things that they are afraid of is that AI might change the statistical data that they give us.”

One of Yseop’s clients told them that they had experienced a dramatic improvement in their report-writing process.

“Reports thousands of pages long that once took four hours to write now take just four seconds with Yseop's AI solution, built on AWS.”

Furthermore, they estimated that in the first year of the partnership up to 10,000 hours of writing and review time were completely eliminated.

While Yseop’s AI dramatically accelerates the drafting process for regulatory documents, human expertise remains indispensable. Dos Santos emphasizes that the AI serves as a powerful tool, but the ultimate responsibility lies with medical writers and domain experts.

“AI excels at processing and organizing data,” explains Dos Santos, “but there are critical elements—like interpreting nuanced clinical findings or weighing the implications of adverse events—that require human judgment and experience.”

The process ensures that every document undergoes rigorous human review before being submitted to regulatory authorities. This dual approach allows users to capitalize on AI's efficiency while maintaining high standards of accuracy and reliability.

“For example, AI might draft a detailed patient narrative,” she continues, “but a medical writer brings context, ensuring the narrative aligns with the broader clinical story and complies with evolving regulatory guidelines.”

This balance is essential for fostering trust. Dos Santos points out, “There are some sections of these documents that will never be written by AI because they demand personal insights and expert interpretation—something no model can replicate.”

By keeping humans in the loop, Yseop ensures that users retain ownership of the content. AI acts as an enabler, but the final decisions—particularly those involving risk-benefit analyses—rest with the human experts.

“The balance will always be that, in order to learn, these models need experts and that expertise comes from humans,” she says.

As well as taking advantage of Amazon Bedrock, Yseop has been using AWS’s cloud infrastructure to bring “all the resources” they need.

“There is a huge amount of data that will be travelling in these models and we have to choose a partner that is as powerful as what we need to use. Security is also key because we can’t really use models where we don’t know what is happening with the data. We can’t take risks.”

It means that the AI models can be deployed in a secure environment, something which dos Santos says “is mandatory for us working with the clients we work for”.

And, she adds, AWS’s reliability and data security is crucial.

“We need to go through a service that isn’t going to fail.”

She is enthusiastic about AI's potential to transform healthcare and how it can enhance human expertise.

“AI can help accelerate access to healthcare, as well as being used in the early phase of drug discovery.”

Long term, as the health industry embraces the benefits that AI can bring, she thinks that AWS will help the firm with its ambitions to grow.

It already counts many of the big pharmaceutical firms among its clients and it is hoping that next year it can offer even more help to them.

Yseop is paving the way for faster medical breakthroughs, ensuring they reach patients at record speeds while driving better outcomes for all.