Researchers have used everyday Fitbit data to train a machine learning algorithm to accurately predict mood episodes associated with bipolar disorder. It opens the door to using a personalized algorithm to drive treatment for the life-affecting condition.

Bipolar disorders (BD) characteristic mood episodes – the extreme swings between depression and mania, followed by a period of remission – can have a huge impact on a person’s work, relationships and health. Treatment of BD is aimed at limiting this impact, which requires prompt identification and treatment of mood episodes.

Leading a new study aimed at finding an accurate way to detect mood episodes in people with BD, researchers from Brigham and Women’s Hospital (BWH) in Boston turned to a now ubiquitous health monitoring device, the Fitbit.

“Most people walk around with personal digital devices like smartphones and smartwatches that capture daily data that can inform psychiatric treatment,” said Jessica Lipschitz, PhD, of BWH’s Department of Psychiatry and the study’s lead author. “Our goal was to use this data to identify when study participants diagnosed with bipolar disorder experienced mood episodes.”

Studies have shown that most people with BD, formerly called manic-depressive disorder or manic depression, experience a change in symptom severity and mood “polarity” at least three times a year. This includes going from feeling very happy, irritable, with a marked increase in activity level (mania), to feeling sad, indifferent or hopeless with very low activity levels (depression). Hypomania is like mania but less severe; it does not cause the impairment in social or work functioning that manic episodes do.

There are two types of BD: bipolar I disorder and bipolar II disorder. BP-I is defined by manic episodes lasting at least seven days (most of the day, almost every day) or mania so severe that hospitalization is required. Separate depressive episodes usually also occur and usually last at least two weeks. Some people with BP-I experience what is called “rapid cycling,” where they have more than four episodes of mania or depression in a year. BP-II is characterized by a pattern of depression and hypomania.

For the current study, the researchers recruited 54 adults diagnosed with BP-I or BP-II and asked them to wear a Fitbit continuously for nine months. The Fitbit Inspire was chosen for its ability to collect data on activity, heart rate and sleep. Participants were also asked to self-report symptoms of depression and mania every two weeks during the same nine-month period.

The data, which included 17 variables such as number of steps, highly active minutes, sedentary minutes, heart rate and resting heart rate, total sleep time, sleep efficiency score, deep sleep duration, REM sleep duration and bedtime, were used to train a predictive machine learning algorithm. The algorithm was able to figure out the importance of each variable in predicting clinically significant symptoms of depression and mania.

The algorithm accurately predicted 89.1% of clinically significant hypomanic or manic symptoms (with a sensitivity of 80.0% and a specificity of 90.1%) and 80.1% of clinically significant depressive symptoms (sensitivity of 71.2%, specificity of 85.6%). Sensitivity refers to the ability of a test to correctly identify patients with a condition; specificity is its ability to correctly identify people without that condition.

The five variables that contributed most to predictions of depression were duration of awakening, total sleep time, median bedtime, resting heart rate, and percentage of sleep in deep sleep. For predictors of mania or hypomania, the top five variables were heart rate, sleep efficiency, percentage of sleep spent in REM sleep, number of highly active minutes, and median bedtime.

“Our results are particularly noteworthy because all input was passively collected, none of the measurements used were privacy invasive, we used common consumer devices, and our methods did not require high levels of Fitbit compliance,” the researchers said. “Other researchers have achieved more accurate mood predictions with more invasive data collection protocols that use data such as geolocation and voice functionswhich can give rise to privacy problems, and textile clothing, which can feel restrictive for patients.”

The findings have the potential to change care models in BD and improve the precision of treatment.

“In the future, our hope is that machine learning algorithms like ours can help patients’ treatment teams respond quickly to new or persistent episodes to limit negative impact,” says Lipschitz.

The study was published in the journal Acta Psychiatrica Scandinavicaand exists as one early view PDF.

Source: BWH via EurekAlert!