Researchers have utilized everyday Fitbit data to train a machine learning algorithm for accurate prediction of mood episodes related to bipolar disorder. This paves the way for using a personalized algorithm to guide treatment of this condition that significantly impacts individuals’ lives.
Bipolar disorder, with its characteristic mood swings between depression and mania followed by periods of remission, can greatly affect a person’s work, relationships, and overall health. Managing this impact requires timely identification and treatment of mood episodes.
In a study aimed at detecting mood episodes in individuals with bipolar disorder, researchers from Brigham and Women’s Hospital (BWH) in Boston turned to the common health-monitoring device, the Fitbit.
Lead author Jessica Lipschitz, Ph.D., from BWH’s Department of Psychiatry, stated, “Most people carry personal digital devices like smartphones and smartwatches that capture data which could inform psychiatric treatment. Our aim was to use this data to recognize mood episodes in study participants diagnosed with bipolar disorder.”
For the study, 54 adults diagnosed with bipolar I or bipolar II disorder were recruited to wear a Fitbit continuously for nine months. The Fitbit Inspire was selected for its ability to gather data on activity, heart rate, and sleep. Participants also self-reported depression and mania symptoms every two weeks during the same period.
The collected data, including variables like step count, active minutes, heart rate, sleep patterns, and more, was used to train a predictive machine learning algorithm. This algorithm identified the importance of each variable in predicting significant symptoms of depression and mania.
The algorithm successfully predicted 89.1% of clinically significant manic symptoms and 80.1% of depressive symptoms. The top variables contributing to depression predictions were sleep-related, while heart rate and sleep efficiency were key for mania predictions.
The researchers highlighted that their passive data collection approach using mainstream consumer devices and non-invasive metrics sets their study apart from others. This has the potential to revolutionize care models for bipolar disorder and enhance treatment precision.
Published in the journal Acta Psychiatrica Scandinavica, the study’s early view PDF is available here.
Source: BWH via EurekAlert!
