Bipolar disorder is characterised by often recurrent episodes of mania and depression. Mania is the defining feature of bipolar disorder. The core symptoms of mania are unusual and persistent elated, elevated or irritable mood and energy, plus further symptoms including increased self-esteem and grandiosity, decreased need for sleep, talking more, having lots of ideas and racing thoughts, being easily distracted, and excessive involvement in goal-directed and/or potentially risky activities. Symptoms of hypomania are similar, but they do not have as much of a severe impact and last no more than four days compared to a minimum of seven for mania.
Mood episodes do not come from nowhere, but we do not fully understand how we can predict them in order to support people promptly and effectively. In between episodes, there are still highs and lows, or ups and downs, in mood, energy, and other symptoms including sleep. The tipping point of when these changes indicate an oncoming episode is an important area of research because identifying early warning signs is a part of therapy for bipolar. Instability in sleep and activity are key triggers and early warning signs (Lobban et al., 2011), making them important targets for therapy (Harvey et al., 2015).
Wearable technologies are increasingly used for tracking risk factors and symptoms in people with mental health difficulties. These are not too invasive and can improve lifestyle behaviours, as previously blogged here.
A recent paper (Ortiz et al., 2025) investigated whether digitally monitored changes to sleep and activity are early indicators of hypomania, and which was the earliest indicator.
Instability in sleep and activity are key triggers and early warning signs in bipolar.
Methods
164 people seeking treatment for bipolar disorder I or II were followed over one year. They rated their own mood using self-report measures of manic and depressive symptoms every week. This also allowed assessment of whether or not a participant had met criteria for a clinically significant hypomanic episode at any point in the study. A wearable device (an Oura ring) automatically recorded sleep and activity in day-to-day life. This included sleep patterns, how long they slept for, how long it took them to fall asleep, and the energy they used.
An Oura ring (wearble device) recorded sleep and activity for a year.
Results
- 50 participants experienced a manic episode at some point during the one-year study.
- Changes to 12-hour sleep variability was the strongest detector of a manic episode.
- When looking at specific manic symptoms, changes to 12-hour variability in activity was the earliest indicator of increased activity (e.g., “I have frequently been more active than usual” or “I am constantly active or on the go at all times.”)
- Changes to day-to-day sleep variability performed well at predicting the manic symptom of decreased need for sleep. These changes could predict an oncoming manic episode up to three days beforehand.
- The sensitivity of sleep and activity changes in predicting subsequent mania decreased as the timeframe of variability widened (e.g., over 12 hours was better than over one week).
Changes to 12-hour sleep variability was the strongest detector of a manic episode.
Conclusions
Sleep and activity changes precede hypomanic episodes by around three days so could be an early indicator for people with bipolar disorder to monitor. The transition to a mood episode can happen quite quickly, so fine-grained tracking of sleep, activity and mood is worthwhile.
Strengths and limitations
This study aimed to address limitations of prior research by using a novel wearable device to track sleep and activity. A year is a long time, so it’s great to see this study lasting this long. It is unusual to have data on sleep and activity for that duration, alongside self-assessments of mood. There was a mix of subjective data (measures that participants completed themselves) and objective data (from the wearable device).
There were many data points and data points matter here for identifying trends and patterns over time. Data was explored for variability rather than absolute change or averaged scores. This gives more fine-grained information and is more like how things are in the real-world.
The measure of (hypo)manic symptoms used to monitor symptoms and identify hypomanic episodes is widely used in research as well as clinical settings, meaning comparisons can be made to other research and it has real world applicability.
People with comorbid sleep disorders were eligible to take part in this study looking at sleep patterns, yet sleep instability may be more likely in people with sleep disorders. However, no comorbidities were assessed or controlled for, and comorbidity is the norm rather than the exception in bipolar disorder, so this likely gives a more realistic picture.
The authors state that not controlling for demographic background is a limitation. There are always biases in research, for example specific people might decide to take part who could differ from those who decline to take part or do not hear about the opportunity to do so. Not controlling for demographics can make it difficult to say what effect this might have had. The sample were mostly working or studying, and well-educated. This is not unusual for bipolar disorder, but it is possible that this sample could be seen as particularly ‘high functioning’. As the current sample were currently treatment seeking, alongside being in work or education, it is possible that people experiencing more social exclusion were not represented. Also, a wearable involves being digitally engaged and the device would also be noticeable. Some people might have been put off for fear of stigma if people asked why they were wearing it.
There are always biases with the self-report of mood, and some evidence that people with bipolar disorder find the specific measure of manic symptoms used difficult for mood monitoring, preferring to create their own questions as reported in this blog. While remote mood monitoring shows promise, it may not be acceptable to all: “people with bipolar are tracking symptoms of mood fluctuation much more specifically and creatively than traditional mood monitoring scales can.” Mood monitoring can also be unhelpful for some people with bipolar disorder (Palmier-Claus et al., 2021), so it would be interesting to know more about how participants actually felt about self-monitoring for a year. Did it perhaps create a fear of relapse or sense of hypervigilance, or was it found to be helpful?
With the current data, I wondered why not look at variability in mood as well, as the mood instability experienced between mood episodes also have an impact on day-to-day functioning. With the weekly ratings, this was possible. Symptoms other than sleep or activity changes could also be early indicators of relapse into a hypomanic episode. An approach such as ecological momentary assessment of mood could have been built in to complement the fine-grained data on sleep and activity, although would have caused more of a burden on participants for engagement with digital technology.
Self reporting of mood is often linked to bias.
Implications for practice
A wearable device, such as the Oura ring, could be useful in practice for sharing with mental health professionals, and in day-to-day life for self-monitoring. So given indications that sleep and activity are early warning signs, this could be a very helpful clinical tool.
While only using self-report for mood has biases, the approach offers real world applicability as weekly appointments for clinician ratings would not be possible. The platform used to collect this data could be linked to mental health professionals.
However, such approaches must be treated with caution (Depp et al., 2016). Those wearing them must be given information to help them to interpret early warnings, and crucially, to know what to do i.e., coping strategies. As above, this is also true of mood monitoring (Palmier-Claus et al., 2021). These approaches could be embedded in existing psychological interventions focused on early warning signs, where there would be support and guidance from a clinician in how to interpret and respond to mood and activity changes (Palmier-Claus et al., 2021).
There are also ethical questions around remote monitoring and where the information goes and how it is used by mental health professionals and services. Alongside further work on their effectiveness for monitoring changes that could be early warning signs of a mood episode, there is a need to gain the perspectives of people with lived experience around the barriers and enablers to wearing devices such as the Oura ring and monitoring their mood, and potentially sharing this with mental health professionals involved in their care.
Mood monitoring could be integrated into clinical care with guidance on how to interpret and respond to mood and activity changes
Links
Primary paper
Ortiz, A., Halabi, R., Alda, M., Burgos, A., DeShaw, A., Gonzalez-Torres, C., … & Mulsant, B. H. (2025). Day-to-day variability in sleep and activity predict the onset of a hypomanic episode in patients with bipolar disorder. Journal of Affective Disorders, 374, 75-83. https://doi.org/10.1016/j.jad.2025.01.026
Other references
Depp, C., Torous, J., & Thompson, W. (2016). Technology-based early warning systems for bipolar disorder: a conceptual framework. JMIR Mental Health, 3(3), e5798. https://mental.jmir.org/2016/3/e42/).
Harvey, A. G., Kaplan, K. A., & Soehner, A. (2015). Interventions for sleep disturbance in bipolar disorder. Sleep medicine clinics, 10(1), 101. https://doi.org/10.1016/j.jsmc.2014.11.005
Levrat, V., Favre, S., & Richard-Lepouriel, H. (2024). Current practices of psychoeducation interventions with persons with bipolar disorders: a literature review. Frontiers in Psychiatry, 14, 1320654. https://doi.org/10.3389/fpsyt.2023.1320654
Lobban, F., Solis-Trapala, I., Symes, W., Morriss, R., & ERP Group. (2011). Early warning signs checklists for relapse in bipolar depression and mania: utility, reliability and validity. Journal of Affective Disorders, 133(3), 413-422. https://doi.org/10.1016/j.jad.2011.04.026
Palmier-Claus, J., Lobban, F., Mansell, W., Jones, S., Tyler, E., Lodge, C., … & Wright, K. (2021). Mood monitoring in bipolar disorder: Is it always helpful?. Bipolar Disorders, 23(4), 429-231. https://pubmed.ncbi.nlm.nih.gov/33570820/
