Biotechnology company Olio Labs on Wednesday published research showing that its preclinical in vivo platform predicts human clinical trial outcomes ranging from gastrointestinal adverse events to cardiac toxicity.

Rodent studies have historically not translated well to human outcomes in clinical trials, while traditional in vivo preclinical studies typically measure just one or two prespecified endpoints that are often shaped by historical convention or researcher bias.

Olio said that its platform leverages state-of-the-art AI vision models trained on expert-labelled data to precisely measure every subtle behavioural change in an animal following drug administration. Behaviours are fed into models trained on human clinical trial data to make predictions.

In the study, a single 24-hour experiment can predict gastrointestinal side effects, cardiac and neuropsychiatric toxicity, and long-term weight loss. Despite being 20 times faster, weight-loss predictions were 70% more accurate than traditional two-to-three-week studies when compared to actual human weight loss in the clinic, Olio Labs said.

Olio Labs also uses the platform to build its own combination therapies, designed to work better with fewer of the side effects that cause patients to quit treatment. The company welcomes additional partners to test the technology.