AI Startups Run Virtual Drug Trials to Predict Which Human Studies Will Fail

A Copenhagen startup, BioinvestGPT, has publicly predicted that drug trials from Biogen and Takeda will fail. Both are due to report results before the end of the year. The company is one of several that run virtual drug trials on simulated patients.

The aim of these companies is to warn drugmakers before they spend years and billions of dollars testing medicines in people. By simulating how a drug might perform in a virtual patient population, they hope to flag studies likely to fail before the human trial is run.

The source text is mostly fragments of page navigation and unrelated headlines, so it gives little beyond this. It does not explain how BioinvestGPT's simulations work, which drugs or conditions the Biogen and Takeda trials involve, what evidence backs the predictions, or how the other companies in the field compare. It also does not say how accurate virtual trials have been so far.

The significance lies in the cost of late-stage failure. Human trials take years and consume enormous sums, so a tool that reliably predicts which studies will fail could change how drugmakers decide where to invest. BioinvestGPT's public calls on two named trials, with results due this year, put its approach to a direct real-world test. If the predictions hold, it would be a notable early data point for AI-based trial simulation, and a Danish company would be at the centre of it.