An international group of researchers has created a digital biomarker aimed at predicting symptoms of depression using data gathered from smartwatches.
The COVID-19 pandemic has triggered a rise in mental health issues. Around one billion people globally deal with various psychiatric disorders. South Korea is particularly affected, with about 1.8 million individuals experiencing depression and anxiety. Over five years, the overall number of people with serious mental health conditions has surged by 37%, reaching around 4.65 million. A collaborative research team from Korea and the United States has designed a technology that leverages biometric data from wearable devices to forecast mood changes and the potential onset of depression symptoms.
On January 15th, KAIST (led by President Kwang Hyung Lee) announced that a research group led by Professor Dae Wook Kim in the Department of Brain and Cognitive Sciences, in partnership with Professor Daniel B. Forger from the University of Michigan’s Department of Mathematics, has developed a system that predicts symptoms related to depression—such as sleep issues, overeating, loss of appetite, and reduced concentration—by analyzing activity and heart rate data collected from smartwatches, particularly for shift workers.
The World Health Organization suggests that an innovative approach to mental health treatment should focus on the brain’s sleep and circadian rhythms, particularly in the hypothalamus, which directly influence our impulsive behavior, emotional reactions, decision-making, and overall mood.
Traditionally, monitoring these internal rhythms and sleep patterns has required extracting blood or saliva every 30 minutes throughout the night to track melatonin levels in the body, along with polysomnography (PSG) tests. These methods necessitate hospitalization and most individuals with mental health issues typically only seek outpatient care, hindering advancements in treatment strategies that consider these critical factors. Moreover, with PSG tests costing around $1,000, mental health care that incorporates sleep and circadian rhythms is often inaccessible to underprivileged communities.
To address these challenges, the use of wearable technology has been proposed as a way to easily gather real-time biometric data, including heart rate, body temperature, and activity levels, without location limitations. However, existing wearable devices only offer indirect insights into necessary biomarkers, such as the state of the circadian clock.
The collaborative team established a method that effectively calculates the shifting phases of the circadian clock through heart rate and activity data obtained from smartwatches. This innovation represents a digital twin that accurately reflects the brain’s circadian rhythm and allows for the assessment of disruptions in these rhythms.
The effectiveness of the digital circadian clock in predicting depression symptoms was validated through cooperation with Professor Srijan Sen’s team at the Michigan Neuroscience Institute and Professor Amy Bohnert from the University of Michigan’s Department of Psychiatry.
The research group carried out an extensive prospective cohort study with about 800 shift workers, demonstrating that the digital biomarker for circadian rhythm disruptions could anticipate not only mood fluctuations for the following day but also identified six key depression-related symptoms, such as sleep disturbances, changes in appetite, reduced focus, and suicidal thoughts.
Professor Dae Wook Kim remarked, “It’s significant to conduct research that uncovers ways to utilize wearable biometric data using mathematics for effective disease management.” He further articulated, “We aspire for this research to pave the way for ongoing and non-invasive mental health monitoring technology, potentially transforming mental health care. By addressing some major challenges faced by marginalized populations in current treatment, individuals might be encouraged to actively seek help when experiencing depression symptoms, preventing crises.”
This study received funding from KAIST’s Research Support Program for New Faculty Members, as well as grants from the US National Science Foundation, the US National Institutes of Health, and the US Army Research Institute MURI Program.
