Have you ever thought that tracking your mental health could be as simple as counting your steps?
Smartphone sensors such as GPS, Bluetooth, Accelerometer and Phone Usage Data could potentially provide a more holistic and objective picture of people’s everyday behavior patterns than traditional assessment methods.
This paper explores the use of smartphone sensors as digital biomarkers, through several studies that bring forth evidence of a strong link between smartphone-tracked behaviors and mental health status. This evidence challenges old paradigms and introduces a more comprehensive approach to understanding mental and behavioral health.
Our findings highlight the potential of digital biomarkers to transform mental health monitoring. This is a crucial step forward in clinical trials and treatment strategies within the medical health sector.
The use of biomarkers in healthcare has exploded over the last few decades, particularly in the fields of oncology and rare diseases. Objective and quantifiable means of diagnosing a disease, monitoring health status, and measuring the effectiveness of new treatments enable the development of more targeted, and consequently more effective and safe treatment approaches, and are the basis of precision medicine.
Until recently however, the benefits of precision medicine - namely better patient outcomes and reduced healthcare costs - have remained out of reach for the mental health field. Mental health disorders have traditionally been diagnosed through episodic, self-reported questionnaires that rely on patients’ recall and subjective descriptions. In order to unlock the same benefits of precision medicine in mental healthcare, it is therefore essential to establish objective biomarkers that can act as accurate, reliable and quantifiable indicators of mental health status.
Daily activity and mobility levels, daily routines, social activity, and even phone usage, as well as numerous physiological features such as heart rate, skin temperature et al, are all influenced by - and therefore can be indicative of - many mental health disorders.
Our understanding of the etiology of mental health disorders is incomplete as it is not possible to identify through brain scans and clinical observation and assessment alone. In addition these methods can pose a burden on the patient to undergo both financially and practically.
However, there are several behavioral and physiological parameters that can be monitored and translated into biomarkers of mental health. For example, daily activity and mobility levels, daily routines, social activity, and even phone usage, as well as numerous physiological features such as heart rate, skin temperature et al, are all influenced by - and therefore can be indicative of - many mental health disorders.
Patients can consent to utilize Digital Health Tools, that demonstrate compliance with current privacy and security standards and regulations, and allow for their behavioral data to be captured in a passive way through the ubiquitous use of smartphones, with embedded sensors, in our daily lives. A wealth of information can be gathered unobtrusively about our daily activities, through GPS sensors, Bluetooth data, and phone usage data.
Within this information are behavioral signals that can potentially be translated into metrics of mental health status, including how much time an individual spends at home; how much time a person spends using their phone, based on phone usage data; and even measures of individuals’ social activity based on Bluetooth data measuring how many other smartphone devices the individual has been in proximity with.
A number of peer-reviewed studies have produced evidence that data gathered from smartphone GPS sensors can reliably indicate mental health status. For example, a group at the University of Virginia1 wanted to investigate the link between emotional state - and more specifically, symptoms of anxiety and depression - and a tendency to spend more time at one’s home.
They collected GPS data from participants’ phones to measure time spent at home, and also several daily self reported mood ratings from participants (via a mobile app), hypothesizing that higher depression and social anxiety symptoms would be associated with a tendency to withdraw, and therefore spend more time at home. The group found a significant association between higher levels of social anxiety, negative mood, and spending more time at home.
People with depressive symptoms tended to visit fewer locations and favored some locations over others, had less regular daily routines, and generally moved less through geographic space.
Similarly, researchers at the Center for Behavioral Intervention Technologies at Northwestern University2 also hypothesized that people’s movement through geographical space would correlate negatively with mental health symptom severity (measured by the commonly used PHQ-9 depressive symptom severity questionnaire).
The study defined a number of behavioral features based on GPS data, including the regularity of participants’ daily movement patterns (circadian movement), how evenly participants' time was distributed across different locations (normalized entropy), and the total amount of GPS mobility (location variance), and found a strong negative correlation between these features and the severity of depressive symptoms.
The findings indicated that people with depressive symptoms tended to visit fewer locations and favored some locations over others, had less regular daily routines, and generally moved less through geographic space.
A compelling body of research underscores the value of GPS data in reflecting mental health conditions. Another recent study3 analyzed data from an EU research program called Remote Assessment of Disease and Relapse–Major Depressive Disorder (RADAR), to explore the relationship between depressive symptom severity and phone-measured mobility (recorded by GPS) over time in 290 participants. The findings again showed a strong negative correlation between depressive symptom severity and phone-measured mobility.
As the above studies demonstrate, there is compelling evidence to support using GPS data gathered from smartphones as an objective, passive and continuous means of tracking mental health status.
While GPS data can be analyzed to infer mental health status by measuring an individual’s movement through space, Bluetooth sensors embedded in smartphones offer the opportunity to measure patterns of social behavior. As mental health has been shown to be closely linked to individuals' levels of social connectedness - or conversely, social isolation - being able to measure social activity objectively and passively is a powerful means of tracking mental health status.
An early framework for bluetooth-based “social sensing” was presented by Yan et al. in 20134 . By analyzing users’ phones’ bluetooth scan records, the researchers were able to count how many other smartphone devices users had been in close proximity to within a certain time period, and from this ‘device count’, infer an individual’s level of social activity.
Monitoring of bluetooth data can therefore lead to the early detection of behavioral changes associated with mental health disorders such as depression, which can help both clinicians and patients to intervene in a timely manner, and prevent more serious symptoms.
A more recent study5 set out to explore whether phones’ bluetooth proximity data, known as the Nearby Bluetooth Device Count (NBDC), could even predict the severity of depressive symptoms. Depressive symptom severity was measured every two weeks, through the commonly used PHQ-8 patient questionnaire, and nearby device proximity data was collected every hour via phone-embedded bluetooth sensors.
The researchers found a number of significant associations between bluetooth measurements and depressive symptom severity. For example, during the two weeks before a worsening of depressive symptoms, participants were seen to have lower total NBDC counts and also lower variety in the devices they interacted with. The study also found that participants whose NBDC sequences were more irregular and chaotic, without a high level of daily periodicity, were more likely to exhibit more severe depressive symptoms.
These changes would be consistent with participants’ reduced social interaction, increased time spent at home, and decreased involvement in activities, which are all behavioral patterns that are commonly observed in depression. Monitoring of bluetooth data can therefore clearly lead to the early detection of behavioral changes associated with mental health disorders such as depression, which can help both clinicians and patients to intervene in a timely manner, and prevent more serious symptoms.
A study by Saeb et al in 2015 found statistically significant differences in phone usage patterns when comparing depressed and non-depressed participants.
While smartphone sensors can gather clinically valuable information about users’ daily lives, such as movement and social interaction patterns, data on usage of the actual smartphone devices themselves can also be revealing of a user’s mental health.
Multiple studies have shown that excessive phone usage, such as spending significant time on social media, or engaging in compulsive phone checking, is linked to higher levels of depressive symptoms7,8. Consequently, researchers have sought to show that tracking phone usage, for example by analyzing the frequency and duration of on-screen time, or number of screen unlocks, can give a strong and accurate signal of a person’s mental health status.
A study by Saeb et al in 20152 explored the relationship between daily-life behaviors - including phone usage, measured by recording the periods of time when the phone screen was on - and the severity of depressive symptoms - measured through PHQ-9 depressive symptom severity questionnaires filled out at the beginning of the study.
They found statistically significant differences in phone usage patterns when comparing depressed and non-depressed participants. Specifically, people with symptoms of depression showed increased frequency and duration of phone usage
Phone usage data is another valuable tool for monitoring mental health status, and can form part of a ‘portfolio’ of digital biomarkers that are able to indicate the presence of mental health conditions much more effectively, sensitively and reliably than traditional assessment methods.
Similarly, Asare et al. conducted a study in 20219 to investigate the validity of phone usage data as a digital biomarker for mental health. They collected data on phone usage, including screen lock / unlock logs, and measured depression severity using the bi-weekly PHQ-8 self reported symptom questionnaire.
By analyzing the variability in phone usage patterns, they found that people were significantly more likely to have higher levels of depression when phone usage patterns were more varied and unpredictable. Moreover, by using machine learning models to analyze a large dataset of phone usage data and PHQ-8 assessments from over 600 participants, the group were even able to predict participants’ depression state based purely on phone usage data, and with over 95% accuracy.
The above studies demonstrate that phone usage data is another valuable tool for monitoring mental health status, and can form part of a ‘portfolio’ of digital biomarkers that are able to indicate the presence of mental health conditions much more effectively, sensitively and reliably than traditional assessment methods.
Academia research has illuminated the potential of passive and objective measurement of mental health through smartphone technology, companies like Healthrythms, Behavidence, and Feel Therapeutics have emerged as pioneers in translating these theoretical theories into tangible clinical tools.
All this academia research has illuminated the potential of passive and objective measurement of mental health through smartphone technology, heralding a transformative era in healthcare delivery. As theories evolve and technologies mature, more and more companies like Healthrythms, Behavidence, and Feel Therapeutics have emerged as pioneers in translating these theoretical theories into tangible clinical tools.
The benefits of passive and objective measurement of mental health status through smartphone technology are evident. Whether through GPS, Bluetooth, or simply phone usage data, smartphones are able to capture information in situ, about daily behavioral patterns, that could otherwise have been missed by traditional, episodic assessment methods.
Companies like Healthrythms use machine learning to translate daily activity into sophisticated insights that can predict bad outcomes before they occur. They are leveraging the power of the smartphones and they passively collect data from smartphone sensors and clinically-validated instruments. Through their platform they run continuous, objective, real-time analysis of patient data so as to better identify, manage, and characterize mental health across a broad range of patient populations.
Additionally, Behavidence through their AI, understands a person’s mobile interactions to help generate a mental health score. Their algorithm calculates a score based on how you interact with your mobile and you can compare yourself to other people diagnosed with ADHD, Depression or Anxiety.
Another company is Feel Therapeutics which integrates smartphone sensors in their platform in order to passively monitor patients. Specifically, their proprietary Digital Precision Medicine Platform (DPMP) facilitates the collection, curation and processing of multimodal data, focusing on the discovery, extraction, and validation of metrics and biomarkers for various use cases in the field of neurology and psychiatry. The DPMP follows the Digital Medicine Society’s DATAcc Framework for Inclusive Development.
Among these data, the platform collects and integrates data available from the sensors of the mobile, including Bluetooth, Global Positioning System (GPS), accelerometer, ambient noise and light data. The continuous data collection along with the data collected through wearables, voice, video, text and clinician’s and patient-reported outcomes, provide a comprehensive 360-degree view of the patient, significantly enhancing our patient monitoring capabilities.
The continuous and passive collection of data, enables us to gather a rich tapestry of information, including daily behavioral patterns captured via smartphone sensors and other modalities. These patterns, encompassing routines, rhythms, activities, and interactions, serve as reliable markers of mental health status.
This holistic approach significantly enhances patient monitoring capabilities and provides clinicians with greater insight into their patients' mental health. Moreover, it empowers patients to monitor their own mental health in real-time, facilitating proactive intervention and support seeking. Thus, leveraging phone-gathered digital biomarkers offers immense potential for more timely and personalized intervention strategies.
