AI Empowers Precision Medicine for Treatment-Resistant Depression

It is often stated that there is an unmet need for more treatments for treatment resistant depression (TRD), but AI is showing that available options can be used more effectively to improve prediction and treatment.

Several studies have been published recently that support the use of AI as a tool for eliminating much of the trial and error in the treatment of patients with major depressive disorders (MDD), including TRD.

Researchers stress, though, that more such studies are essential because AI is rapidly being integrated into treatment for MDD, and TRD and its utility needs to be validated.

An example of the expanded use of AI or machine learning in TRD was the October 22 acquisition of Options MD by New York-based Resilience Lab, a mental health treatment group that says Options MD’s proprietary AI clinical tool gave it the confidence to expand into difficult-to-treat forms of depression.

“Options MD has shown impressive clinical outcomes with a 27% remission rate within just three months of beginning treatment, versus an industry benchmark of 14%, and a 65% retention rate at three months, versus the industry benchmark of 30%,” the practice claims.

Resilience Lab is a treatment provider available to covered patients in New York, New Jersey, Connecticut, Massachusetts and Pennsylvania.

Similarly, Potomac Psychiatry in Rockville, Maryland, has also launched an AI tool it calls “Dr. Holo” to help patients with TRD.

The group uses its AI model to personalize care for patients and reduce haphazard treatment assumptions that lead to failure.

AI is almost as new on the scene as practices’ utilization of it in TRD and MDD, but the evidence suggests this technology might be ready for prime time.

A 2022 study evaluated the perceived usefulness of a clinical decision support system (CDSS) named Aifred, using psychiatrists and personal care providers who employed the CDSS aid in prediction and treatment for anonymized real-life patient examples.

The authors said 60% of physicians found it to be useful in treatment selection, with the greatest acceptance coming from family physicians, who tend to feel less confident in treating serious depression.

Myriam Tanguay-Sela

“Moreover, 50% of physicians would use the tool for all patients with depression, with an additional 35% noting they would use the tool for all patients with severe or TRD depression,” wrote lead author Myriam Tanguay-Sela, with Aifred Health Inc., of Canada, and her co-authors.

Importantly, over several sessions, the physicians’ confidence and trust in the AI tool increased significantly. They indicated it tended to give them a stronger rationale for the chosen treatment course.

“Forty percent of physicians made reference to the benefit of being able to communicate the motivation for a treatment decision to the patient, ‘especially to give patients concrete numbers about their remission probabilities,’” the authors wrote.

Investigators state that AI can improve the accuracy of disease prediction by crunching volumes of existing clinical data on patients. Depressive disorders are often caused by an underlying disease, which an AI model could suss out.

A South Korean study found that an AI tool was 88.% accurate in tracking and predicting the course of depressive disorders. “AI can predict not only depressive disorders but also other mental conditions, such as suicidal tendency and postpartum depression with high accuracy,” wrote lead author Yoonseo Park, M.Sc., of the Department of Convergence Healthcare Medicine at Ajou University, Suwon, South Korea.

A separate study looked at the accuracy of predictive models specifically in TRD and found that one model achieved 62% accuracy for predicting this form of MDD.

The authors concluded AI can improve early detection of TRD, help with development of personalized treatment plans, prioritize care for patients at risk, and inform research and treatment guidelines.