Mental illness is invisible, but it is a disease that many people already suffer from. Approximately 1 billion people worldwide are affected by various mental illnesses. In South Korea, the number of patients with depression and anxiety disorders is around 1.8 million, and the total number of individuals with mental illness is estimated at about 4.65 million. This represents a 37% increase over the past five years. Experts say we are in a pandemic situation regarding mental illness.
Another issue with mental illness is the difficulty of accurate diagnosis. Diagnosing and predicting symptoms related to depression, such as sleep disorders, feelings of depression, loss of appetite, overeating, and decreased concentration, requires costly hospital diagnosis.
To measure endogenous circadian rhythms and sleep states, it is necessary to draw blood every 30 minutes overnight to measure changes in melatonin hormone levels in the body and conduct polysomnography (PSG). If hospitalization is unavoidable and there is no insurance, the examination expense alone exceeds 1 million won. Socially vulnerable populations find it difficult to diagnose and treat mental illness.
Recently, researchers from the Korea Advanced Institute of Science and Technology (KAIST) and the University of Michigan developed a technology to predict symptoms related to depression using only a smartwatch, eliminating the need to visit a hospital.
Professor Kim Dae-wook from the Department of Brain and Cognitive Sciences at KAIST and Professor Daniel Folger from the Department of Mathematics at the University of Michigan announced on the 15th that their research team developed technology to predict symptoms such as sleep disorders, feelings of depression, loss of appetite, overeating, and decreased concentration from the activity levels and heart rate data collected from smartwatches.
Smartwatches can easily collect various biometric data, such as heart rate, body temperature, and activity levels, in real time without spatial constraints. However, smartwatches only provide indirect information and were unable to predict actual symptoms related to depression from the data.
The research team developed technology to accurately estimate the daily changing phases of the biological clock using heart rate and activity data collected from the smartwatch. The biological clock plays a role in maintaining the 24-hour rhythm within the suprachiasmatic nucleus located in the pituitary gland. It informs our bodies of the current time of day, facilitating the secretion of necessary hormones, and allows us to sleep and wake up at regular times. When the rhythm of this biological clock is disrupted, symptoms related to depression, including sleep disorders, may occur.
The technology developed by the research team succeeded in creating a digital twin that accurately describes the circadian rhythms in the brain based on data collected from smartwatches. They used this to accurately predict when depressive symptoms would appear in response to external factors disrupting the circadian rhythm. The research team validated whether the technology works properly with 800 experimental participants. As a result, they successfully predicted a total of six symptoms, including next day's mood and typical depressive symptoms such as sleep problems, appetite changes, decreased concentration, and suicidal thoughts.
Professor Kim Dae-wook noted, “It is very meaningful to conduct research that provides clues for applying wearable biometric data, which has not been well utilized, to actual disease management using mathematics. This should resolve the issue where socially vulnerable individuals need to take proactive actions, such as contacting counseling centers when experiencing depressive symptoms to receive help, thereby proposing a new paradigm for mental health care.”
The research team plans to file a patent related to the technology developed this time and aims to collaborate with a U.S. wearable data analysis company to proceed to actual medical device development.
Reference materials
npj Digital Medicine (2024), DOI: https://doi.org/10.1038/s41746-024-01348-6