A predictive machine learning model was able to identify how effectively patients with Crohn’s disease would respond to long-term usage of ustekinumab, according to a new study in JAMA Network Open.

Researchers from the University of Michigan trained the model using demographic and laboratory data obtained in previous clinical trials. The data described 401 patients with active Crohn’s who had been monitored while taking ustekinumab, an immunosuppressive drug, over the course of at least 42 weeks.

Once trained, the model was able to reliably identify how a Crohn’s patient who had been given one dose of ustekinumab would react to the treatment in the long term. The technology could therefore be used to predict those patients who may not respond well to biological monotherapy, saving them money and reducing delays in achieving remission.