JMIR Publications

ABSTRACT

Background:

When trying to intervene early to prevent manic episodes, clinicians are hindered by patients underreporting their symptoms as they are becoming hypomanic or manic (i.e., when these symptoms are emerging). Continuous passive monitoring using wearable technology offers a potential solution to address this problem by capturing behavioral signals that do not depend on self-awareness. However, its clinical utility requires validation before implementation.

Objective:

This study aimed to evaluate whether passively collected wearable data can detect hypomanic symptoms using an explainable machine learning framework. We applied explainable machine learning to densely sampled multimodal data from a two-year observational study of digital monitoring in 178 patients with a diagnosis of bipolar disorder (BD) (mean follow-up 540 days; range 120-706 days). Data on activity and sleep were collected continuously using wearable technology. Self-report data was collected daily on their mood, energy, and anxiety; and weekly using the Altman Self-Rating Mania (ASRM) scale. To identify robust digital markers of hypomanic symptoms, we analyzed 252 time- and frequency-domain features from 43 participants who experienced at least one ASRM-defined symptomatic interval (defined as an increase in ASRM score of ≥10 for a week).

Methods:

We applied explainable machine learning to densely sampled multimodal data from a two-year observational study of digital monitoring in 178 patients with a diagnosis of bipolar disorder (BD) (mean follow-up 540 days; range 120-706 days). Data on activity and sleep were collected continuously using wearable technology. Self-report data was collected daily on their mood, energy, and anxiety; and weekly using the Altman Self-Rating Mania (ASRM) scale. To identify robust digital markers of hypomanic symptoms, we analyzed 252 time- and frequency-domain features from 43 participants who experienced at least one ASRM-defined symptomatic interval (defined as an increase in ASRM score of ≥10 for a week).

Results:

Passively collected data achieved a high accuracy, comparable to daily self-reported data in detecting symptomatic intervals (AU-ROC 0.94 ± 0.02 vs 0.92 ± 0.06; p>0.05). Combining wearable and daily self-reported data yielded a comparable high performance (AU-ROC 0.95 ± 0.02). The high classification performance based on wearable data was primarily explained by reductions in spectral entropy and flux across domains, indicating decreased complexity and variability in mood, activity, and sleep patterns, consistent with a shift toward pathologically stable states.

Conclusions:

Passive monitoring using wearable technology can detect hypomanic symptoms with high accuracy, supporting its potential as a clinically meaningful tool for continuous, objective monitoring in patients who underreport hypomanic symptoms. The clinical usefulness of passive monitoring with wearable technology in patients with BD needs to be confirmed in a large prospective study.