The COVID-19 pandemic has had a significant effect on our country’s most vulnerable populations, from who gets the SARS-CoV-2 virus to its impact on mental health and access to care. For example, the pandemic has tripled the already large numbers of people with depression, increased anxiety and substance abuse and has added to the backlog of wait lists for appointments and treatment.
On the flipside, the pandemic has positively transformed healthcare by unleashing new digital health solutions, such as telehealth and remote monitoring.
As I scan the horizon for medical technology to help members identify and evaluate innovative emerging technologies, I’ve seen an explosion of digital technologies in the behavioral health space. The development of these technologies has the potential to transform the behavioral health care paradigm and increase access to underserved populations; but it’s not without challenges.
As a point of reference, it’s been estimated that there are more than 20,000 different digital mental health products. It’s a large, broad field, with the vast majority of products being mobile phone, app-based wellness products that promote mental well-being or do things like provide coaching, affirmations and guided visualizations.
There are also more advanced digital technologies that fall under the category of “digital phenotyping.” These are a step up in complexity. In addition to doing some of the same things that the wellness products do, they begin to capture information about the users’ mental state. For example, one of these products might capture how you’re feeling during the day along with your sleep quality, exercise habits and diet and then use that data to identify potential problems related to your mental health.
There are also products at the high end of the digital phenotyping category which act similar to more conventional medical technologies. For example, an app might use a chatbot to present a scientifically validated mental health assessment questionnaire that it then uses to make a diagnosis. There are some chatbots like this that’ve been used for screening and diagnosis of things like anxiety, stress, depression and post-traumatic stress disorder.
Prescription digital therapeutics actually go another step beyond and provide some sort of psychological therapy. So, these might use a chatbot for the conversational element to deliver some core principles of cognitive behavioral therapy. If/when these digital therapeutic products become widely available, they could add a lot of value and truly change the behavioral health paradigm.
The impact of digital technologies
Digital technologies could provide behavioral health options for individuals in rural communities where there may not be a therapist anywhere nearby. They also could provide access for those in cities who may lack transportation or the ability to pay for services.
Digital technologies in the behavioral health space could be a way to help individuals who cannot or do not seek help because of the perceived stigma of mental illness. They could also provide an option for an individual who experiences a middle-of-the-night panic attack when you can’t just call your therapist. Interestingly, some literature shows patients may actually be more likely to open up and tell things to a chatbot that they might not even tell their therapist.
These technologies also could address workforce issues such as the severe shortage of mental health care professionals. And one of the biggest potential advantages is that the technology is so easily scalable. Once you’ve developed the software, you can distribute it to as many users as you want without much increase in operating costs.
The ability to improve access to underserved populations and scalable use can go a long way to improving the disparities in access to behavioral health care, but it’s not without challenges.
Challenges with digital technologies
In practice, there are still a few bugs to work out with this technology. For example, chatbots are not yet very good at interpreting things like hyperbole and sarcasm, which may make them seem inhuman. This can lead to a high dropout rate, with some studies finding more than 90% of users quit using these apps within about the first month.
Chatbots also can get things wrong. When evaluating for depression, for example, a mistake that leads to a missed major depression diagnosis can lead to a very dangerous outcome for a patient. That’s why, just like for other medical devices, you need to have rigorous clinical trials that prove safety and efficacy, you need to have FDA approval for a well-defined indication, and you need good patient selection criteria.
As these technologies work to achieve FDA approval, hopefully we should start to see more rigorous evidence development.
Another fundamental challenge with this technology is that it relies on the self-disclosure and truthfulness of the user. So, while human questioners can often subjectively judge the truthfulness and completeness about what they’re hearing, computers aren’t so good at reading between the lines.
In addition, some data in the literature notes that diagnostic accuracy may be lower when patients report their own symptoms compared to when a therapist enters the patient symptoms. This suggests a potential weakness in the underlying programming in that it has some trouble translating common speech into technical concepts.
Even with all these challenges, many of which are not unique to digital technology, given the limitations in access to mental health services, the true comparator may not be an app versus a therapist, but the app versus nothing at all. So, these technologies may have a real advantage in getting at least some care to people who may not otherwise get any care.
About the author: Joe Cummings, PhD., is Vizient’s technology program director within the Performance Improvement Collaboratives group. Dr. Cummings' areas of expertise include technology assessment, evidence-based medicine and clinical supply integration and he has research experience with a broad array of medical devices, including artificial intelligence, remote monitoring, robotics, rapid diagnostics, enhanced biomaterials, metagenomics and telehealth. His educational background is in biomedical engineering.