More than 21% of U.S. adults experience depression, greatly impacting their quality of life. Many people with mild‑to‑moderate depression can improve their symptoms by adjusting daily habits like sleep, exercise, diet and social interaction, according to Jyoti Mishra, Ph.D, associate professor of psychiatry at University of California, San Diego School of Medicine. However, because depression is highly variable between people, a one‑size‑fits‑all lifestyle approach isn’t very effective.
In a first-of-its kind study, Mishra and her team developed a machine-learning guided lifestyle coaching program based on data collected via personal devices about participants’ mood and daily habits. They found that participants who implemented the program experienced significant reductions in depressive symptoms after six weeks. The findings offer a promising approach for remotely delivering personalized depression treatment tailored to each individual’s circumstances. The study was published in NPP – Digital Psychiatry and Neuroscience.
During a two-week period, 50 adults with mild‑to‑moderate depression wore a smartwatch that tracked heart rate and exercise levels. In addition, they logged their mood and answered short questions up to four times per day about their quality of sleep, diet, activity level, and how often they talked with friends or family.
The team developed a machine learning model unique to each participant based on this data to discover which lifestyle factors best predicted an individual’s low moods. Then, each participant worked with a health coach to implement an individualized mood augmentation plan, or iMAP.
“Our goal was to figure out the top lifestyle factors driving the depressed state, which would be different for different people, and to find out if by targeting those factors through personalized coaching, people would actually feel better,” said Mishra, who co-directs UC San Diego’s Neural Engineering and Translation Labs (NEATLabs).
For the next six weeks, the participants worked with their coach to implement their iMAPs.
“Every person in the trial was doing different behavioral therapies that are already well-established in the literature depending on their top predictive factor,” said Mishra. “Some people were working on a cognitive behavioral therapy program for insomnia, others on maximizing the kinds of physical activities they were already doing in their daily lives, enhancing social connections, or implementing a healthy mood diet-based intervention.”
After working with the coach via short video calls for six weeks, participants:
Reported a large reduction in depressive symptoms; 55% of participants no longer had depression after the treatment as measured by the Patient Health Questionnaire-9 (PHQ-9), a standardized depression screening test.
Reported a 36% drop in anxiety symptoms as measured by the Generalized Anxiety Disorder-7 (GAD-7) screening test.
Reported significant improvements to their quality of life.
Scored higher on brief memory and attention tests.
What’s more, the researchers found that the treatment effect persisted during the three months they continued to follow the participants after the intervention ended.
“Clinical trials show that most current interventions only show about a 30% benefit on average in terms of depression remission; here we see a near doubling of that due to targeting the top lifestyle predictive factors with data-driven personalized coaching,” said Mishra.
Mishra thinks the intervention may be more effective because it is a departure from generic recommendations for behavioral health.
“Everybody knows that we should eat healthier diets or try to sleep eight hours or exercise 150 minutes per week and so on,” she said. “But I think personalized insights can be more empowering than these general guidelines because they’re not so overwhelming. When one is in a depressed state, it's not possible to change everything about one's life — you're just trying to survive and function on a day-to-day basis.”
Though the study was small, it provides the first evidence that digital monitoring, machine learning-derived insights and brief, personalized weekly coaching delivered remotely may be a promising integrated approach to address mild-to-moderate depression in large groups of people. A larger, controlled study of this personalized therapeutic approach is needed to validate the findings.
Read the full study here.
Additional co-authors on the study include: Jason Nan, Suzanna Purpura, Satish Jaiswal, Houtan Afshar, Vojislav Maric, James K. Manchanda and Charles T. Taylor at UC San Diego; and Dhakshin Ramanathan at UC San Diego and VA San Diego Medical Center.
The study was funded in part by a seed grant from the Hope for Depression Research Foundation.
Disclosures: Taylor is a paid consultant for Neuphoria Therapeutics (Bionomics), atai Life Sciences and Engrail Therapeutics, and receives payment for editorial work for UpToDate. Other authors declare no competing interests.
