Medical Opinions Are Often Divided, but Tech Can Bring Them Together

Patients place their faith in medical professionals for making sound clinical decisions based on their diagnoses. But what if these decisions vary significantly from physician to physician—who can patients trust to make the final call?

Machines may offer the best answers, say Canadian mental health researchers who recently published a study featured in the Journal of Applied Behavior Analysis. In their work, the team, led by Marc Lanovaz and Kieva Hranchuk, say a machine learning platform they developed could soon be ready to enter the clinic and support physicians in making better patient recommendations.

According to Lanovaz, medical professionals inconsistently prescribe behavioral interventions for treating mental health conditions. Moreover, they also frequently disagree on the effectiveness of behavioral remediation practices. Ultimately, this affects patients, with many receiving inadequate treatment to meet their mental health needs. 

Lanovaz and Hranchuk spearheaded a study that compiled simulated data from a cohort of over 1,000 participants receiving behavioral therapies. The researchers then put man against machine, comparing treatment conclusions generated by professional behavior analysts against those derived from a machine learning platform that the researchers developed.

The results demonstrated two ways in which the computer-based system outperformed the doctoral-level behavioralists. Firstly, the AI-generated results were more consistent, with the professionals only coming to similar conclusions about 75 percent of the time. In addition, machine learning was less prone to errors than professionals.

As the next steps, the researchers plan to integrate their algorithms into an easy-to-use app that would assist physicians in making data-driven decisions on the best path forward for their patients. The researchers say this technology is by no means intended to replace doctors but instead to make clinical decision-making more precise and consistent.

“For example, doctors could someday use the technology to help them decide whether to continue or terminate the treatment of people with disorders as varied as autism, ADHD, anxiety, and depression,” commented Lanovaz.