How Data-Driven Therapy Is Changing Mental Health Treatment

By Anna Codrea-Rado, Contributor

“Let’s bring it back into the room” is a classic phrase therapy patients have been hearing for decades. In the mental health treatment of tomorrow, however, that “room” may be a digital one.

Text therapy and virtual clinicians are becoming more prevalent, delivering treatment in new, easier-to-access ways. Behind these innovations are big data sets which are transforming how mental health treatment is executed. Here, mental health experts explain how this data-driven therapy is more accessible, accountable, and results in better patient treatment.

CBT 2.0

Cognitive Behavioral Therapy (CBT) was first developed in the 1960s and is now widely accepted as an effective treatment for a range of mental health disorders, such as anxiety and depression.

CBT is a skills-based approach to treatment, typically delivered in a short course of six to eight weeks. In these sessions, a therapist will identify patients’ triggers and provide them with the tools to remodel the way they think or behave.

While CBT is one of the most widely-used forms of therapy, in practice, only about half of the patients achieve full recovery. A recent study by the University of York in the U.K. found that over half of CBT users relapsed within 12 months. Another study that compared the use of CBT and counseling in depression treatment found that half of all patients, regardless of the type of intervention, did not show reliable improvement.

“In general, we don’t have a very good system of quantifying the effectiveness of mental health treatment,” said Ashley Womble, author of “Everything Is Going to Be OK.” “Unlike medical interventions, there are no blood tests that can measure the success of therapy.”

But that’s where data comes in.

“It was through the data that we discovered that individual therapists can get into the 80 percent recovery rate.”

—Valentin Tablan, senior vice president at Ieso

Ieso Digital Health is a platform that provides CBT via a text-based online conversation between a therapist and the patient in a secure virtual therapy room. Commonly known as text therapy, a number of other companies, including BetterHelp and TalkSpace, offer a similar service. By the end of 2018, Ieso had treated 30,000 patients and logged 180,000 hours of therapy. Using anonymized data from the text conversations, the company has been able to improve recovery rates for its patients.

“It was through the data that we discovered that individual therapists can get into the 80 percent recovery rate,” said Valentin Tablan, senior vice president of artificial intelligence at Ieso.

Ieso built a deep learning model that’s able to identify different ideas that are discussed in a therapy session. By running the model over 90,000 hours of therapy, the company has been able to correlate when a therapist delivers a certain type of content in a session with an improvement in the symptoms of the patient.

For example, Ieso’s data analysis has shown that recovery rates are likely to increase if the patient has been given homework, such as mindfulness exercises to practice, by his or her therapist, or if the two of them have set an agenda together. “The patient needs to be involved in their own recovery,” Tablan said.

“We now have a data-driven model of what actually helps patients get better,” he added.

Therapy for All

Data-driven therapy may not only improve patient recovery outcomes, but also widen access to treatment.

“Tech is democratizing mental health by making it more accessible for everyone,” Womble explained. “The old school method of therapy, where you make an appointment weeks in advance and travel to see a therapist in his or her office for 45 minutes, just doesn’t work for everyone.”

At the Institute for Creative Technologies at the University of Southern California, researchers are looking into whether the solution to offering more accessible therapy resides in virtual therapy. Dr. Skip Rizzo, the institute’s director of medical virtual reality, and Gale Lucas, research assistant professor, have developed a “virtual therapist” called Ellie.

“People prefer opening up to her instead of a human,” Lucas said. “They disclose more to the virtual therapist than a real therapist.”

Ellie was designed as a screening tool to help identify if a veteran is at risk of PTSD or depression by interpreting behavioral indicators. She was stationed in Veterans Affairs buildings as a testing ground. “The system isn’t just listening, it’s quantitatively looking at their behavior and predicting their likelihood to have PTSD,” continued Lucas.

Despite Ellie not being a “real” therapist, patients respond well to her in a large part because she’s a piece of technology. “People feel safer with a machine,” Rizzo said. “They aren’t being judged by it.”

Connected Mental Healthcare

As connected devices become ubiquitous, mental health experts expect to see data play even more of a role in treatment. “As the face of therapy changes, we will see more of an influence of big data analysis and AI extraction,” Rizzo noted.

As he explained, there are currently only three data points a clinician can analyze in a therapist context: what patients self-report, their physiology, and their observable behavior in the clinical setting.

Lucas noted the difficulty these data points can present in the physical setting. “It’s very hard when you’re self-presenting to the therapist,” he said. “There’s a barrier there, plus the therapist is just getting a little snippet of someone’s life just once a week.”

The next phase for data-driven therapy will be measuring health data when the patient is out in the real world, in between sessions. For example, using existing wearable technology that collects and stores health data, mental health professionals may be able to determine risk factors for a person’s well-being.

“To the extent that we feel comfortable using wearables, there is so much opportunity for ongoing monitoring,” Lucas explained. “If we can get data from facial expression, vocal tone, how much someone exercises—all these factors are rich, big data to [look into] predictors of PTSD and depression.”

Data analytics is revolutionizing science across the board. As this extends to mental health treatment, it will mean better access to care for those who need it the most. As Tablan concludes, “We can look at the data and understand therapy much better than [ever] before.”