BioPharma, Pharma, Artificial Intelligence

AI startup Unlearn adds $50M for better, faster, smaller & cheaper clinical trials

Unlearn.AI’s Series B round follows a positive European Medicines Agency draft opinion finding that the company’s artificial intelligence technology can be used in Phase 2 and Phase 3 clinical trials. The startup is now on the hunt for pharmaceutical partners that want to use this “digital twin” technology to speed up their clinical research.

 

Throughout the history of medicine, drug developers have tried to make clinical trials faster and more efficient. Artificial intelligence and machine-learning methods are some of the newer approaches to quicken the pace but these technologies run into long-held beliefs that speed lowers the bar for evidence, said Charles Fisher, CEO of Unlearn.AI. The startup now has findings from a regulator stating otherwise.

The European Medicines Agency recently issued a draft opinion stating that Unlearn’s technology can improve the speed and efficiency of randomized, controlled clinical trials. Now the San Francisco-based company has something more. On Tuesday, Unlearn unveiled $50 million in financing to support its plans to bring its technology to a wide range of clinical trials.

Unlearn’s approach employs “digital twins,” which are virtual representations of something in the real world. Using historical patient data crunched by the company’s AI technology, Unlearn computes a twin for every patient in the trial. By populating the control arm with these twins, a clinical trial requires fewer patients, which eases the enrollment bottleneck that are big contributors to the time and cost of these studies. Fisher says the technology makes clinical trials better.

“A better clinical trial is something that is faster, has lower cost, but delivers the same evidence,” he said. “You require fewer patients to get the same evidence.”

Unlearn has been discussing its digital twin approach with the FDA and the EMA for more than two years. As of now, technologies used in a study are reviewed as part of the clinical trial protocol for a drug. But Fisher said that his company wanted to go beyond that to get a more thorough regulatory stamp on the technology.

The EMA issued a draft opinion on Unlearn’s technology late last month. The regulator said that the application of this technology leads to efficiencies while still maintaining the same level of rigor as a randomized controlled study. The agency wanted to know whether analysis of a drug could be biased by incorporating AI or machine learning. Fisher said that Unlearn was able to mathematically prove that Unlearn’s technology does not introduce bias. That evidence led the agency to conclude that the technology could be used in the
primary analysis of Phase 2 and 3 studies.

“We think it’s a huge, important step for us as a business and for the broader industry,” Fisher said. “The question has always been, can you apply machine-learning methods with historical data to do clinical trials that are more efficient, successful, and regulators accept—and provide the same level of evidence as traditional randomized controlled trials. Our argument has been for a long time, yes of course.”

The new financing follows another milestone for Unlearn. In February, Merck KGaA was the first pharmaceutical company to sign on as a partner to the startup. The multi-year alliance will focus initially on using the digital twin technology in the development of the German company’s immunology pipeline. The alliance may also expand to other therapeutic areas. Financial terms were not disclosed, but Fisher said that Unlearn will receive milestones tied to the progress of Merck drugs developed with the digital twin technology. Now that the startup has EMA validation for its technology, Fisher said he is looking to strike up similar partnerships with other pharma companies. While pharma companies are the main focus, Fisher added that Unlearn is also willing to work with the contract research organizations that run many clinical trials.

The initial focus of Unlearn was applications of its technology to clinical trials for neurological disorders, such as Alzheimer’s disease, multiple sclerosis, and amyotrophic lateral sclerosis. With the new funding, Fisher said that Unlearn is now looking to expand to other therapeutic areas. Specific disease areas may depend on finding a partner, because Unlearn builds its model using data about a disease and those data are often in the hands of a drug company.

Unlearn will also continue to develop and improve its technology. Fisher aims to have his company build more models that are more predictive of disease progression. These models could incorporate more kinds of data, such as images and genetics data. By making models that are more predictive, Fisher said that clinical trials can become even smaller. Achieving those goals will address pharma company interests in making studies faster and lower in cost while also meeting the evidentiary requirements of regulators. These advances also appeal to patients.

“We’re trying to make the control group of the trial as small as possible,” Fisher explained. “By making the control trial smaller, more patients get access to the therapy, which is why most patients participate in clinical trials.”

Photo by Flicker user mike goehler via a Creative Commons license