Health tech developers must work with front-line clinicians when developing artificial intelligence tools, a new paper argues.
Published last week in Frontiers in Psychiatry, its authors aimed to create a framework for practical AI implementation that addresses the benefits and challenges for mental health clinicians. To date, few resources exist on this topic, they argue.
The paper focused on three categories of emerging technologies: automation, engagement and clinical decision support tools. Automating repetitive tasks helps alleviate workloads, while engagement tools help keep patients on the right track between sessions. Decision support tools, meanwhile, can help clinicians detect mental illness earlier. Ultimately, the paper's authors told Fierce Healthcare, the goal is to advocate for AI as a supplement, not a replacement, to clinicians’ expertise.
“For clinicians, documentation is just a crushing burden,” co-author Katherine Kellogg, professor of management and innovation at MIT, told Fierce Healthcare as an example. AI is an obvious intervention to help mitigate burnout, she added. “We believe that these AI technologies have tremendous potential to change clinicians’ quality of life.”
Importantly, AI tools can also support clinicians in training and accommodate various skill sets—a critical benefit at a time when there is an enormous demand for and shortage of therapists, she added.
Addressing uncertainties and challenges related to AI implementation should be up to clinical leaders across healthcare organizations as well as health tech developers, the paper argues.
And it's critical to include clinicians in the development process of AI tools. “These technologies are often opaque, black boxes,” Kellogg said. “The key voice that is missing from this conversation is the voice of the frontline clinician.”
In her practice as a clinical psychologist, co-author Shiri Sadeh-Sharvit, Ph.D., has observed a lack of awareness about AI and its uses among her colleagues, she told Fierce Healthcare. Sadeh-Sharvit is also chief clinical officer at Eleos Health, a clinical app for voice AI that transcribes and summarizes therapy sessions. Approximately 20% of the company is made up of therapists in training or practicing therapists, Sadeh-Sharvit noted, and the company prioritizes working with clinicians to develop and update its technology.
“Help is on the way. There are tools already available to clinicians to use,” Sadeh-Sharvit said. “I do want to assure therapists that we are not going to replace them; we want to add tools to help them do their job better.”
This review was the first step in providing clinicians a guide for the uses, benefits and challenges of using AI on the front lines. But further research is needed—the kind that tracks clinicians’ actual implementation of AI in the field, Kellogg and Sadeh-Sharvit said. And, while behavioral health is an area lagging behind other healthcare settings in tracking and implementing AI, “we have not solved this problem in any of the domains,” Kellogg noted.