New research into why high doses of steroids don’t work for some severe asthma patients used machine learning techniques to discover a dozen clinical variables that predict who those patients are, reports the Pittsburgh Post-Gazette.

The research was led by Wei Wu, a faculty member in Carnegie Mellon University’s computational biology department, working with longtime collaborator Dr. Wenzel as senior author.

“We hope that approaches like this will lead to improvement in the way physicians use prednisone and not over-treat patients,” Dr. Wenzel said. “If a little isn’t doing better, more won’t. We’re left trying to help physicians identify patients who will get better.”

Ms. Wu said that once software is available, doctors will be able to predict how their patients will respond. If a patient falls into the most resistant subgroup, they can try other treatments.