Digital Psychiatry: How Startups are Turning Everyday Tech into Tools to Fight Mental Illness

October 30, 2017
Contributed by: Jenny Hsu

Mental healthcare in America – an unmet need

According to the most recent (2015) statistics from the National Institutes of Mental Health, 1 in 5 American adults (43.8 million) have been diagnosed with a mental illness. A subset (9.8 million) experiences a serious mental illness that substantially interferes with life activities. Experts estimate that mental health is a $150 billion dollar industry, which consists of mental health providers (psychiatrists and therapists) and chemical drugs.

The demand for treatment has exceeded the supply in the last few years, with the gap predicted to become even larger as the stigma toward mental illnesses decreases and more people seek treatment. Currently, 20.3% of adults with a mental illness – about 9 million adults – are seeking treatment yet face barriers to getting the help they need. The scarcity of mental health providers is partially to blame; there is only one mental health provider for every 529 individuals, and the disparity is even sharper within specialties.

In addition, big pharma has withdrawn resources for the R&D of psychiatric drugs by 70%, despite this sector of the market being historically among their most profitable. In the New Yorker, Dr. Richard Friedman, Professor of Clinical Psychiatry at Weill Cornell Medical College, proposes three reasons for this withdrawal: (1) psychiatric drugs are essentially copies of each other or of their predecessors, with no improvements on efficacy and no new targets; (2) clinicians lack animal models and biomarkers to measure drug responses; and (3) most importantly, we simply do not know enough about the brain. As a result, only 8% of drugs designed to work on the brain achieve clinically significant benchmarks, compared to 15% of other drugs.

The evolution of the digital therapist

Your brain on digital drugs

Digital interventions for mental illness are technologies that work through the same psychological methods that a therapist uses in an office. Most digital interventions employ cognitive-behavioral therapy (CBT) – they focus on challenging patients’ thoughts and beliefs, their “cognitive processes.” These arise from cell-to-cell connections in the brain regions that regulate emotion, motivation, and informational processing. In mental illness, dysfunctional connections generate a distorted, distressing view of the world. The connections are then strengthened over time, amplifying unhealthy moods and behaviors. Patients are therefore trained to modify their thoughts, beliefs, and behaviors to weaken these connections, resulting in positive changes in emotions.

In digital format, CBT may involve a program that reminds patients to perform a breathing exercise when stressed, or to rewrite a self-criticism in a more realistic light. The program might also allow patients to enter their symptoms and receive treatment suggestions, and to track their progress in relation to treatments and life events. Thus, the digital format allows constant and almost universal access to healthcare. The best websites (MoodGym, Beating the Blues, Fear Fighter) are already the standard of care in some countries.

Most importantly, CBT is the most widely accepted approach in mental health care, having amassed support in the form of over 300 effectiveness trials. Neuroimaging studies show that practicing CBT decreases activity in the brain regions responsible for emotionality, and increases activity in those responsible for thoughtfulness. Delivering CBT and other psychotherapies in a digital format could therefore be the first instance of modifying an organ without any physical input to the body. In some instances this approach might even be preferable to drugs. “Let’s say that you’re 30-something years old and you live until you’re 80-something… that’s a lot of pills to swallow,” says Dr. Yan Leykin, Associate Professor in the Clinical Psychology PhD program at Palo Alto University, “but therapy has lasting effects after you stop seeing a therapist, because you actually learn something.”

What excites investors and psychiatrists even more than the science is the rate at which this technology is advancing and spreading. Smartphone ownership in mentally ill populations meets the national average (up to 77% in 2017 from 35% in 2011), and most digital interventions now exist as smartphone apps. Dr. Craig Barr Taylor, Professor Emeritus at the Department of Psychiatry and Behavioral Sciences at Stanford University, refers to these apps as the “second generation” of digital interventions. MIT Technology Review writes that they are understood in Silicon Valley as “the third phase” of medicine, “the successor to the chemical and protein drugs we have now, but without the billion-dollar cost of bringing one to market.”

Navigating the hype

In the latest study from the IMS Institute for Healthcare Informatics (2015), the number of health intervention apps exceeds 165,000. Mental health apps have by far the largest share of this pie at 29%. Research from Dr. Stephen Schueller’s team at Northwestern University has shown that few therapists actually prescribe digital interventions because they do not have time to navigate this broad and complex landscape. However, IMS reports that adoption by providers can increase the customer retention rates of health-related apps by up to 10%.

Dr. Joaquin Anguera, Assistant Professor in Neurology and Psychiatry at UCSF, argues that we have reached a critical point when digital interventions must be held to “a higher standard, as high a standard as most drugs are, if not higher.” This will allow individuals to “better understand how to interpret and digest what a meaningful result is.” Anguera echoes the sentiments of many mental health experts: that digital interventions must be subject to a peer-reviewed standard if they are to be seriously considered as an alternative to chemical drugs.

Moving forward – areas for innovation

Building a support network

Dr. Taylor argues that to provide more informed treatment, integration with the healthcare system should become a distinguishing feature of most second-generation apps.  Currently, only 2% of all health apps connect to healthcare providers. Pacifica is a prime example of the integration that apps need. Designed for anxiety, Pacifica allows users to log their habits, take various breathing and muscle relaxation exercises, and set daily goals for managing their moods. Early this year, the developers released a new feature, Pacifica for Clinicians, which allows mental health providers to access their patients’ activities and provide feedback.

In addition, connecting patients with each other increases the rate of recovery, as reviewed by a team at the Dartmouth Institute for Health Policy and Clinical Practice. Of the most downloaded health intervention apps, 65% connect to social media, and the number of health intervention apps that connect to social media has increased from 26% to 34% between 2013 and 2015.

To be effective, interactions must be structured and moderated, a key research finding that the messenger app Koko takes into account. A Koko user can write posts of stressful situations to be sent anonymously to other users, who are prompted to offer empathy, identify negative thought patterns, and reframe the sentences to be more positive. While waiting for a response, the user can also serve as a respondent in the system, rewriting the sentences of other users. Importantly, a bot screens both the posts and the responses for offensive, off-topic, or unhelpful content. Its developers received $2.5 million in August 2016 to extend the artificial intelligence behind its screening methods to virtual assistants like Siri and Alexa, so that these might be trained to deliver empathetic responses to trigger words.

Making the medicine taste good

Taylor notes that people have traditionally used digital interventions for only a very short period of time. Dropout rates are indeed similar in web-based self-help conditions to in-person treatments and can be as high as 83%, with an average of 31%. Therefore, Taylor argues that second-generation apps can and should distinguish themselves by improving engagement.

Dr. David Mohr and his team at Northwestern University have shown that patients remain engaged if they can access multiple treatment options. The app that they developed, Intellicare, is unique in that it supplies 13 “mini-apps” and encourages users to try new apps each week. After its first year on the market, the number of active users, app launches per user, and hours of sustained usage were higher for Intellicare than for any previous app.

Headspace is another successful venture, boasting over 15 million users. Users can access many different guided meditations that fall under five different improvement areas (e.g. “health”), each with multiple subcategories (e.g. “stress”), or one-off meditations based on an activity (e.g. “running”). As an added incentive to complete exercises, the app also provides virtual awards that can be viewed in a profile page.

Looking ahead, the ultimate goal is to deliver the most beneficial interventions for a particular person based on past preferences, severity of symptoms, and the user’s schedule and current environment. This capability is akin to adjusting the difficulty, or “dosage,” of a digital intervention, which Anguera argues is much faster and less risky than with a chemical drug. It also requires integrating features that unobtrusively measure whether a patient’s condition has improved, regardless of whether the patient continues to use the intervention.

New biomarkers for mental health

A team in Switzerland has shown that synchronizing a digital intervention to a sensor that detects biomarkers of mental health improves compliance. Currently, only one in ten health-related apps has the capability to connect to a sensor, and the majority of these are fitness trackers and not mental health-related. However, analysts predict that the global market for wearable sensors will grow more than 88% between 2016 and 2020, and companies are quickly mobilizing.

Australian-based Medibio shapes its approach on the link between mental health and measurable changes in the autonomic nervous system. Using these measurements, Medibio has predicted depression and PTSD accurately in two large studies. Other groups, such as Sonde Health in collaboration with the MIT Lincoln Lab, seek to identify “vocal biomarkers” of mental health – vocal features like tone, pitch, rhythm, rate, and volume – so that clinicians might someday track patient progress remotely by analyzing speech samples.

Some initiatives focus on having the smartphone double as a sensor, eliminating the need to purchase and carry an additional tool. This approach is more accessible to the general population, since only 2% of the U.S. has a wearable device, and the use of wearables is higher among those already motivated to monitor their health. Smartphone users provide a wealth of data – GPS signals, accelerometer activity, proximity to other Bluetooth devices, frequency of text messages – translate into predictions about location, fitness, and social interactions. Taken together, these predictions paint a picture of the user’s mental state.

Of these initiatives, Mindstrong Health has recently gained the most publicity. They attracted former director of the National Institute of Mental Health, Thomas Insel, to take on a co-founder role in May 2017, and raised $14 million in Series-A funding in June 2017. Mindstrong aims to assess mental health through “metadata” such as patterns of typing, scrolling speed, or the repetition of key words and phrases. By the end of 2017, it will reveal the findings of its first clinical trials.

Putting digital interventions to the test

According to Dr. Josef Ruzek, Co-Director of the Center for m2 Health and a Professor in the Department of Psychiatry and Behavioral Sciences at Stanford, the uptake of an app depends not only on its features and user friendliness, but also on its evidence base. He states that the skill sets required for developing an app are not the same skill sets required for evaluating it, opening up new opportunities for third party testing organizations. Dr. Taylor adds that apps are traditionally tested in randomized trials “comparing A with B, to achieve some desired outcome, so energies are focused on trying to make users use the product.” As a psychiatrist, Taylor is more interested in identifying interventions that work robustly, which requires analyzing effect size in large cohorts.

In addition to apps in development, apps on the market are now beginning to be tested in clinical trials. Pacifica is being tested by the University of Minnesota, and Headspace by two independent groups at Stanford University and NUI Galway. Intellicare has launched an ongoing national research study and reports significant decreases in depressive and anxiety symptoms in the 96 participants who have already been tested. Overall, the number of trials focused on mental health-related apps increased by 32% between 2013 and 2015.

Furthermore, as with a chemical drug, testing with well-designed placebos has been linked to the credibility of the product. One interesting example is, which aims to help users sleep better by training them to combat negative thoughts. It was initially tested against a placebo app that featured exercises like visualizing neutral objects or mundane tasks. Another is Koko, the aforementioned messenger app. Before being developed into a commercial product, it was tested at MIT Media Lab with a placebo that had users practice creative writing instead of actively challenging negative thoughts.

Reaching new markets

Because the course of a mental illness can depend on social and environmental factors, developers are beginning to think about how to design digital interventions for specific ethnic groups. These apps need to overcome language and cultural barriers, and to be tested in populations that are underrepresented in large clinical trials. “What we’re realizing is that when we have this idea that technology is all we need and can do everything without the person, people are not as interested,” says Dr. Adrian Aguilera, Assistant Professor at the UC Berkeley School of Social Welfare and a clinical psychologist by training. His team develops digital interventions for Spanish-speaking communities, but he mentions an interest in other groups, such as Asian Americans, for whom the stigma of mental illness has been shown to be a primary deterrent for seeking treatment. For the 2.2 million Asian Americans who suffer from mental health issues, “technology may be a way to access content about mental health without having to go see somebody,” Aguilera adds.

Niche markets also include specific mental illnesses. Apps for anxiety disorders and depression take up 18% of mental health apps apiece, dominating the market. In contrast, there is only one app for schizophrenia – FOCUS, which is not yet available on the market. A team at Dartmouth University found that most patients easily used FOCUS to manage their symptoms over months, even during the most high-risk periods following hospital discharge, suggesting that schizophrenia patients are able and willing to adopt digital interventions. In addition, Dr. Taylor credits the broad user base of PTSD Coach, the only digital intervention for PTSD, in part because “anxiety, stress, depression, and wellness apps are the most common, but PTSD is more complicated, so this particular app met a niche that others did not.”

“Keeping at it”

As of now, experts can only speculate on the potential of digital interventions to be a cure for the most severe of mental health conditions. Although Dr. Leykin cautions that some cases cannot be managed without medication or a therapist, he also mentions that “most of what can be done therapeutically could be computerized.” And as the ability to create attractive and convincing user interfaces continues to improve, app developers will keep pushing the boundaries.

Any surge of innovation will bring with it new questions. For example, what is the best way to assess demand in a user base that, while open to new treatment options in the best of times, struggles with hopelessness and lack of motivation in the worst? In addition, as digital interventions become more advanced, would they represent a risk to patients if used improperly, and should the FDA step in to regulate them?

Despite these uncertainties, Dr. Aguilera remains optimistic about the contribution of digital interventions. “One of my interests is thinking about how to merge technology with people, and finding the best sweet spot,” he says. “At the individual level we’ve gotten a lot of very positive feedback on how this is helping them [our patients] make positive change on a daily basis and feel more connected. I think that gives us motivation to keep at it.”

 Jenny Hsu is a Science Communications Fellow with BCBA and a graduate student in Genetics at Stanford University with a focus on stem cell biology.  She has previously worked in psychiatry research at Massachusetts General Hospital.