A Novel Unsupervised Machine Learning Approach to Assess Postural Dynamics in Euthymic Bipolar Disorder
Summary
People with bipolar disorder (BD) show altered posture even when euthymic. Reduced postural variability during the day is linked to higher illness burden, particularly more depressive episodes.
Area of Science:
- Neuroscience
- Psychiatry
- Biomedical Engineering
Background:
- Bipolar disorder (BD) is characterized by mood fluctuations, including euthymic, manic, and depressive states.
- Motor and postural abnormalities are recognized during mood episodes but less studied during euthymia.
- The relationship between postural dynamics in euthymia and overall illness burden in BD remains underexplored.
Purpose of the Study:
- To investigate postural abnormalities in euthymic individuals with bipolar disorder.
- To explore the association between these postural changes and various measures of illness burden.
- To understand how illness burden impacts posture and sleep consolidation during euthymia.
Main Methods:
- Collected 24-hour posture data from 32 euthymic participants with BD using a wearable sensor.
- Extracted nine time-domain posture features and performed unsupervised clustering.
- Analyzed associations between posture features and 12 clinical characteristics of illness burden.
Main Results:
- Participants clustered into three groups based on daytime, evening, or nighttime postural dynamics.
- Higher illness burden correlated with reduced postural variability, especially during daytime.
- Frequent nighttime postural transitions and upright posture were linked to a higher number of lifetime depressive episodes.
Conclusions:
- Euthymic individuals with bipolar disorder exhibit distinct postural abnormalities.
- These postural changes are significantly associated with illness burden, particularly the history of depressive episodes.
- Findings highlight the impact of illness burden on posture and sleep consolidation during stable mood periods in BD.
