Arshya Vahabzadeh talks at Exponential Medicine

Mental health is a global health crisis, but there are 3 key ways to tackle this, according to Arshya Vahabzadeh, MD, Harvard Medical School faculty of psychiatry.

Among those 3 keys are improvements in business & cultural factors, developments in computer science & informatics, and advancements in neuroscience & genomics.

Vahabzadeh spoke at the recent Exponential Medicine conference on how trends in technology cross-pollination with other fields like urban design—and even self-driving cars—could help reduce the severity of mental health.

The annual Exponential Medicine conference brings together over 500 entrepreneurs, organizations, innovators, and academics to discuss the future of medicine. These topics covered the latest technologies, including nanotechnology, digital manufacturing, data & machine learning, blockchain, artificial intelligence, robotics, biotechnology, mobile technologies, and virtual reality.

Speakers from the University of California, San Francisco (UCSF), New York University (NYU), Dell Medical School, Arizona State University, and the University of Pennsylvania also shared their thoughts on topics ranging from the process of innovation to changing current clinical practice, global health, medical education, entrepreneurship, and XPRIZE.

Watch the whole 15-minute presentation:


How mental health affects the world

Vahabzadeh asserted that mental health is a significant problem on the global, national, and individual levels. Mental health makes up a large proportion of the global disability burden, from adolescence throughout adulthood and into the elderly, according to the World Health Organization. This, in fact, is the leading cause of disability in the United States—with the highest cumulative disability-adjusted years. Suicide, in particular, is among the top ten causes of death in the U.S.

And, when people are held back by mental illnesses, this translates to trillions of dollars of economic output. In China and India, for example, their lost output from 2012 projected through 2030 will be greater than all other categories of chronic health conditions, based on research by the National Bureau of Economic Research in 2013.

The problem: Only 28.7% of U.S. adults receive any sort of depression treatment. The good news is that people are using different perspectives to understand and tackle these issues—from a business and cultural perspective.

Compounding this is the ambiguity of diagnosis, Vahabzadeh says. Take depression, for instance. Typically, the diagnosis rests on a person’s mood history and symptoms. With the current diagnostic criteria, these depressive symptoms have 227 different ways of combining that could lead to a diagnosis of major depressive disorder.

One of my colleagues from the University of Netherlands looked at the seven most common depression research scales,” Vahabzadeh says, “and what he found was 52 different symptoms, but rarely do symptoms show up in all of the scales, and only six appeared in all the scales that he looked at, and 21 were only in one scale! So, what are we measuring?


Business & cultural factors

From Couch to Cloud: Technology 4 PsychiatryMental health clinicians typically think of symptoms, says Vahabzadeh. Neuroscientists, on the other hand, look at these illnesses through circuits of the brain. Technology experts also look at mental illness through other systems beyond the typical healthcare system—political, legal, and environmental. And startups are looking at products that could help reduce mental illness.

These different lenses and viewpoints can be used to come up with new solutions for mental health. For instance, there are an increasing number of apps and startups. These apps have great potential for digital detection of symptoms, collecting data such as how fast one is typing, how one is speaking, and how fast a person is moving. Vahabzadeh predicts that many of the singular apps that have offered only one function—such as GPS tracking for friends—will combine into larger systems for multimodal monitoring.

These startups can be ushered by healthcare technology accelerators. The accelerators help startups by providing a virtual community of other founders linked with funding and potential customers. One such accelerator, NeuroLaunch, bills itself as the world’s first neuroscience startup accelerator with 11 portfolio companies since 2014.

And, video games and entertainment—systems that are not typically linked with healthcare—can be employed in the form of an adaptive video game to assess attention deficit hyperactivity disorder (ADHD). Akili Interactive Labs, which uses technology developed at UCSF, published about how their video game could perform better than standard cognitive measures.


This cross-pollination also applies to medications from the pharmaceutical industry being paired with smartphone apps. Pfizer, for instance, released Moodivator to track goals and mood symptoms. Pear Therapeutics has also created a web interface with a smartphone application to help track medication compliance.

Vahabzadeh also predicts that autonomous vehicles can contribute to mental well-being. Increased commute time—in a 2014 Office of National Statistics report—leads to a decrease in happiness, an increase in anxiety, and a decrease in life satisfaction. However, the emergence of self-driving cars has the potential to impact alcohol use, affect substance use, reduce distracted driving, and decrease stress into what Vahabzadeh bills as “therapeutic pods.”

Similarly, urban design and mental health intersect. Instead of prioritizing the fastest route to a destination, Vahabzadeh asserts that urban planners, as well as mapping services, could create routes that are the least stressful or even cater to specific mental conditions. Persons with autism are sensitive to stimulation, and a route could be designed to have the lowest sensory load. Persons with substance use or alcohol use issues could use a route with the least triggers to relapse.


Computer science & informatics

Quantitative Mental Status ExaminationWithin the realm of computer science, technologies—such as image matching, motion tracking, and voice analysis—are becoming commoditized into commercially-available application programming interfaces (API’s) online from Google, Microsoft, and IBM Watson. Some of those API’s include Google Cloud Vision API, Microsoft Azure Cognitive Cloud services, and IBM Watson’s personality assessments. Academic research is moving into commercial services, such as Sonde and Cogito, that use computerized speech and language analysis.

Schizophrenia has correlations with speech patterns of phrase lengths and semantic coherence. These can be detected by speech algorithms to predict psychosis onset in high-risk youths, from that of patients with schizophrenia, according to Bedi and Gillinder.

Computer science could also help train therapists. In a 2016 paper in Transactions of the Association for Computational Linguistics, Althoff, Clark, and Leskovec used computational analysis of counseling conversations. Their team analyzed 408 counselors, 15,555 conversations, and 663,026 messages. With the help of computers, the authors found that more successful counselors had three key attributes: They were able to clearly identify the problem, spend more time in problem-solving, and learn to change the course of a conversation if the conversation was not going well.

Similar analyses can be conducted on medical hospital discharge notes to help predict suicide risk. Positive valence words—such as “good,” “social,” “positive,” and “consistent”—were shown to correlate with a 30% decreased risk of suicide, in a 2016 JAMA Psychiatry paper by McCoy et al.

Virtual Reality in PsychiatryVirtual reality (VR) and augmented reality technologies are also becoming more mainstream. These technologies allow people to experience an immersive world with visuals driven by computers. Work at the University of Southern California has immersed soldiers in VR exposure therapy environments to help with post-traumatic stress disorder (PTSD). A 2016 paper by Freeman et al. looked at immersing users in a crowded London public transport subway to help with delusions.

Startups, such as Vahabzadeh’s Brain Power startup, is starting clinical trials of its wearable system for autism. The device encourages children to make eye contact and detect facial expressions. Overall, Vahabzadeh asserts that VR has promise in psychiatry, as VR can help assess symptoms, determine environmental issues as causes, and help with treatment.

Another groundbreaking development is the communication of facial expressions. Facebook, this year, introduced the Oculus Rift, which allows users to not only interact in the same virtual space, but communicate by raising eyebrows and mouth expressions using the hand controllers. This has implications for group psychotherapy, with PTSD & mood triggers, and with OCD exposure therapy.

Machine learning technologies also will play a greater clinical role in the future. These technologies can help predict someone’s risk level and response to treatments. In a 2015 JAMA Psychiatry article, Kessler et al. found that such machine learning predictions can help identify who is at most risk of suicide. In 2016’s The Lancet Psychiatry, Koutsouleris et al. found that machine learning can reliably predict treatment outcomes in first-episode psychosis who is most impaired for first-onset psychosis. The journal also featured work by Chekroud et al., finding that machine learning can predict how well patients respond to a particular antidepressant—citalopram—with more accuracy than psychiatrists.


Neuroscience & genomics

Rise in Psychiatric Genetic FindingsFinally, learning about how mental health is tied to physical health—as well as the invisible building blocks of life—will influence psychiatric practice. Rao discusses in Primary Psychiatry how many organs and systems, such as the lung, the heart, gastrointestinal, autoimmune, and endocrine—which includes hormones and blood sugars—can be affected by and influence depression.

In fact, there have been recent studies on a gut-brain connection, via the “gut microbiome”. The gut microbiome refers to the bacteria that makes up one’s intestines. This microbiome can generate neurotransmitters and metabolites that influence how the brain functions.

Vahabzadeh says that the process of “genotyping”—or looking at a person’s genetic code—can help make prescriptions safer and more effective as well.

There is tremendous amount of overlap in the genes that we’re finding,” Vahabzadeh says. “A third of the genes [found in a research study] were shared by multiple disorders. Another example, in this case, five different mental health disorders were sharing the same 15 genes.

Steven Chan, MD, MBA (@StevenChanMD)
Dr. Chan is a Clinical Informatics fellow at UC San Francisco (UCSF)'s Division of Hospital Medicine, serving as editorial boardmember for the Journal of Medical Internet Research (JMIR) Mental Health, and develops cutting-edge research in the areas of digital mental health, with applications for cultural psychiatry and underserved minority health. Steve's ideas, thoughts, and research have been featured in JAMA, Healthcare, JMIR (Journal of Medical Internet Research), Wired, PBS, and NPR Ideastream. Steve serves as Vice Chair for the Workgroup on Mental Health & Psychiatric Apps at the American Psychiatric Association (APA), a part of the Committee on Mental Health Information Technology.


All comments are moderated. Please allow at least 1-2 days for it to display.