AI aims to improve telehealth
Collaborative project develops virtual assistant to facilitate mental health care
Posted on: November 17, 2021; Updated on: November 17, 2021
By Michael Ettlemyer, ettlemye@mailbox.sc.edu, 803-777-9757
Kaushik Roy knows firsthand the importance of access to health care.
“On a personal level, I have faced the issue of high demand and not enough care,” he says.
It’s a problem that existed before the pandemic — especially in rural and underserved areas — and has been exacerbated by COVID-19 over the past year-and-a-half. Roy’s experience is part of the motivation for him to see the completion of a project aimed at using artificial intelligence to facilitate care between mental health patients and providers.
“It’s super important for this project to materialize,” says Roy, a doctoral student in computer science at the College of Engineering and Computing. “The impact is going to be huge. For me, it’s very important to get it done so that more people can realize the potential of it.”
The COVID-19 pandemic created challenges to the quality of care not only by increasing the number of patients needing mental health treatment, but also by moving many appointments and communications from in-person to telehealth. In many cases, active monitoring of patients stalled as clinicians have struggled to keep up with demand.
A clinician-assistive and patient-supportive tool that would passively monitor the patient’s mental health conditions and actively inform clinicians about changes in patients’ conditions would improve the telehealth or in-person visit. That is where work being done at the University of South Carolina makes its impact.
The health setting is going online, and the pandemic accelerated that.
Kaushik Roy
The collaborative project brings together students at the University of South Carolina’s AI Institute led by Amit Sheth, founding director of the institute; and medical experts from the College of Nursing (Phyllis Raynor, Sara Donevant and Ronda Hughes), Prisma Health (Dr. Meera Narasimhan, Suzanne Hardeman) and more. All have the goal of improving access to mental health care and streamlining communication between patient and provider.
Researchers sourced data from active conversational public health forums on platforms such as Reddit, generating a database on mental health conversational data that spans 11.5 million interactions between users from January 2020 to February 2021.
Algorithms were designed to function as a virtual health assistant that makes clinicians aware of a patient’s condition. Experiments conducted with extensive involvement from the College of Nursing further developed the tool and have further demonstrated its potential clinical usefulness.
Armed with this new algorithmic tool, clinicians can help understand what is happening with their patients between visits. The process looks like this: A patient expresses symptoms and behaviors during a clinical visit. Then, the virtual assistant leverages this information to bootstrap its ongoing conversation with the patient between visits. The patient benefits from the conversation during this time, and the clinician is better informed oft he patient’s details during the next visit.
“The health setting is going online, and the pandemic accelerated that,” Roy says. “Before, AI set reminders to take medication or track health vitals. Now, because of circumstances, there is increasing need for AI that can actually assist the provider or therapist, help with knowledge of the patient and help understand it. There is a big need and huge potential to use AI in that way.”
Sheth says the collaboration with mental health professionals has been key to developing the initial technology. Funding proposals are pending, and additional funding will allow them to expedite testing, eventually leading to clinical trials.
“Qualified clinical help in mental health is in severe short supply,” he says. “While several mental health conditions will continue to need the attention from a clinician, there is a significant opportunity to help patients with milder cases using such assistive technology.”
Having seen the stresses that the pandemic has put on those around them, Roy and his fellow students understand the need.
“I already know this system definitely can benefit a large section of people,” Roy says.