Executive Summary

• Whereas physical health is measurable (“a healthy heart rate ranges from 60 to 100 bpm”), mental health has no universal “scores” or quantifiable parameters

• Measurable and personalized mental health biomarkers can, however, be a powerful tool to support diagnoses and guide patients and treating physicians along the journey of healing

• The system of tomorrow needs to be objective, continuous, and proactive, all of which the newly emerging sphere of digital phenotyping combines

• Digital phenotyping, which refers to data collected from personal devices such as smartphones, is used by multiple groups (from consumers to B2B partners). We see a powerful tool that can pave the way toward clinically useful markers that can refine diagnostic processes, tailor treatment choices, and improve condition monitoring for actionable outcomes.

Model for a scalable, integrated multi-user platform for digital phenotyping research

Balancing Physical and Mental Wellness

Access to healthcare is becoming increasingly democratized. With a couple of clicks on our phones, we have an approximate understanding of what our bodily symptoms might mean. A couple more clicks and a (virtual) visit to the doctor is scheduled. Especially in recent months, the ease of digital healthcare has developed for both physical and mental health. However, there is a missing link: whereas physical health is measurable, mental health overall is not.

Certain scales for assessing mental well-being consider indicators such as demographics, health history, and social and occupational status. However, there are no universal “scores” or quantifiable parameters in place, such as for physical health, where we, for example, know that a healthy resting heart rate ranges from 60 to 100 bpm.

Correlation of physical and mental health

Somatization occurs when psychological (or emotional) factors express themselves as physical (somatic) symptoms, e.g., somatic anxiety includes chest pain and fatigue. Although somatization may occur, these indicators play out differently in different people. That is why the need for personalized measurement of mental well-being is needed.

Typically, mental well-being has been reliant on self-reported and subjective judgments. However, mental health is much more individual and personal than physical health. Measurable and personalized biomarkers can be a powerful tool to support diagnoses and guide physicians and treat patients along the journey of healing. There is an emerging field of research and companies developing ways to uncover measurable mental health insights. One unique way of doing this is digital phenotyping.

“Building The Thermometre for Mental Health” (credit to the 2018 paper by Insel and Chauvin)

Mental healthcare today tends to be:

• Subjective,

• Episodic,

• Clinic-based, and

• Reactive.

The system of tomorrow needs to be:

• Objective,

• Continuous,

• Ubiquitous, and

• Proactive.

Today, several digital approaches are implementing these factors of what is considered the future of mental healthcare. It is happening in various ways: apps using visual biomarkers via a laptop or smartphone camera allow you to measure physiological and mental health parameters from faces, such as London-based thymia; auditory solutions enable computer-generated voice support, such as Berlin-based Clare&Me.

Beyond these approaches lies a wave of continuous data collection startups which aim to understand the sources of distress. One way of achieving this to help with diagnosis and related treatment interventions is digital phenotyping. It is a concept that observes online proxies of behavior and emotions to understand the human psyche and dimensions of mental health care. In thisnovel area of science, research uses data from smart devices to build a rich, personalized digital picture of behavior, track markers of e.g. depression, anxiety, and other conditions, and develops new ways to diagnose illness, choose effective treatments and detect relapse before it occurs.

How Big Is Digital Phenotyping?

Digital phenotyping is a nascent market that lies at the fringes and crossover of many existing health fields. It can transform models of care in new and unseen ways, for example, by providing a common ground to map patients’ physical and mental health as they navigate a disease state or monitor their mood when faced with a new lifestyle or diet.

Because digital phenotyping touches on multiple areas of healthcare all at once, its market size is difficult to quantify. We have instead provided a visual summary of its positioning within the broader healthcare space:

Digital phenotyping and the delivery of healthcare supported by new infrastructures and tools may be particularly synergistic: as discussed in one of our previous articles, hybrid care is a rising force, disrupting traditional care delivery and contributing to a potential £256 billion Digital Health market by 2027. Similarly, the increasing ubiquity of wearable technology, especially medical-grade devices, within the general population opens the possibility of partnerships and usage of devices and digital phenotyping.

The potential of digital phenotyping has been well documented in the scientific literature, especially within mental health. It is sometimes heralded as the answer to a scalable model of care to meet the surging demand for mental health services. However, digital phenotyping has gotten a negatively connotated reputation in some regard due to concerns of social media companies infringing privacy while collecting data. The effective and privacy protected application from research to clinic is the step needed to shift public opinion from an underhand advertising tool to a force of societal and healthcare good.

Target Data and Persona

Generally, both active and passive traced data of smartphone usage is taken into account for digital phenotyping. These data inputs include:

• Electronic Behaviour (e.g. social media interactions and kinesthetics including tapping patterns)

• Auditory and Visual Input (e.g. using a smartphone camera to detect facial microexpressions)

• Sensors (e.g. phone accelerometer for location and mobility, and wearables for physical biomarkers).

Digital phenotyping data inputs

However, what data tells and the actual state of mind of users may only sometimes be in sync. That is why a combination of active (e.g. prompting users to input a daily score of their mood) and passive data collection is needed to bridge the gap between the mental health picture painted digitally and in real life. Then, potential causation factors can be detected and met with relevant interventions. An example is the effect and reduction of social media on users’ moods.

Business Models: A Look at the Start-ups Landscape

That being said, who is the audience here? Can digital phenotyping be used as a consumer tool, as a research tool, or as a tool for clinicians? The answer is all of the above, depending on the use case.

We have identified incredible research applications using digital phenotyping, e.g. the impact of social interactions on chronic diseases and associated preventative health interventions or how sleep affects concentration and satisfaction levels. Passively traced data will unlock the most extraordinary doors for health research. As per the consumer segment, we see a smaller section of people who may actively track how their perceived state of mind correlates with what data tells them. We see a lower stickiness factor here, as self-motivated nudges do not usually yield excellent adherence levels.

Digital Phenotyping for Mental Health: Condensed Startup Landscape

Instead, we see that the mode of psychological treatment is shifting from reactive to proactive: this means that not only patients but also treating clinicians perceive the need to intervene before a relapse of e.g. a depressive phase occurs. There is a powerful niche of psychotherapists adopting digital phenotyping to continuously check in on their patients and comprehend continuous developments according to data beyond individual sessions. However, clinicians still hesitate to adopt such intervention support tools: speaking with various experts in the field has shown us that professionals might be open to adopting new operating systems in their practices but less so when it comes to changing the way they treat patients. The journey of including both patients and clinicians in a digitally-measured mental health journey is still in development.

As digital health consultations become a norm, we expect that the digital phenotyping segment is yet to proliferate in the years ahead: there will be the tangible value provided by non-invasive information about patients, allowing clinicians to understand an individual’s mental state better and plan for adequate treatment. Yet for this to occur, the actual access to professional mental health care needs to ease, for example through group therapy or B2B mental healthcare solutions.

Mental healthcare, as well as physical healthcare, is still anchored by the human element of a professional. However, the (digital) quantification of a patient’s health through data may result in clinically useful markers that can refine diagnostic processes, tailor treatment choices, improve condition monitoring for actionable outcomes, such as early signs of relapse, and develop new intervention models.

We are excited about the rise of these models and look forward to connecting with startups in the space!

Sources

https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8709510/

https://www.blackdoginstitute.org.au/research/data-analytics/digital-phenotyping/

https://sharpbrains.com/blog/2018/12/04/on-mental-health-and-the-advent-of-digital-phenotyping

https://www.nature.com/articles/s41746-019-0166-1