Investigators assessed if data from fitness trackers could effectively identify mood swings in individuals with bipolar disorder.
Research shows that fitness trackers can help identify mood changes in bipolar disorder to enhance treatment.
Researchers from Brigham and Women’s Hospital, part of the Mass General Brigham healthcare network, studied whether data gathered from fitness trackers could reliably identify mood phases in people diagnosed with bipolar disorder. Their research, published in Acta Psychiatrica Scandinavica, reveals that it is indeed feasible to pinpoint times when bipolar patients are experiencing episodes of depression or mania with high precision using information from fitness tracking devices.
“Most individuals carry personal digital gadgets like smartphones and smartwatches that record daily data that may be useful for psychiatric care. Our aim was to utilize this information to recognize when participants with bipolar disorder were undergoing mood changes,” stated Jessica Lipschitz, PhD, the lead author and a researcher in Brigham’s Psychiatry Department. “We aspire that in the future, machine learning models like ours could assist treatment teams in responding quickly to new or persistent episodes, minimizing their adverse effects.”
Bipolar disorder (BD) is a long-term mental health condition characterized by severe mood fluctuations, including periods of depression, mania, and hypomania, which are followed by times of stability. Identifying and addressing new and ongoing mood episodes is crucial to reduce the influence of BD on patients’ lives. While earlier studies have shown that personal digital devices can successfully detect mood episodes, they have not utilized approaches designed for widespread clinical application.
As an implementation scientist, Lipschitz and her team concentrated on methods that could be broadly applied in clinical environments. They specifically utilized commercially available personal digital devices, minimal data filtering, and entirely passive, noninvasive data collection. By employing a novel kind of machine learning algorithm, they achieved an 80.1% accuracy rate in detecting significant symptoms of depression and an 89.1% accuracy rate for significant symptoms of mania.
The researchers assert that “overall, these findings advance the field toward personalized algorithms suitable for all patients, rather than just those who show high adherence, have access to specialized devices, or are willing to provide invasive data.” Their next objective is to implement these predictive algorithms in everyday care, allowing them to enhance the management of BD by alerting healthcare providers when their patients are experiencing depressive or manic episodes between regular appointments. Additionally, the team is working to extend this research to encompass major depressive disorder.
