Linking Activity Patterns and Mood States in Bipolar Disorder: A Longitudinal Case Study based on Actigraphy Signals

Background Mood fluctuations in Bipolar Disorder are closely linked to changes in activity levels, sleep quality and daily rhythms. Therefore, actigraphy could be a valuable tool in the investigation of such mental health conditions, aiding in understanding, diagnosing and treating such disorders. It is a crucial problem in the healthcare of bipolar patients to find objective features, e.g., in diurnal or nocturnal motion patterns, which can promote prediction of sudden state-changes of the patient. To this end, we carried out a comprehensive mathematical analysis of an extremely large set of actigraphy recordings (spanning through more than 600 days) of a bipolar outpatient. Results The research employed cutting-edge statistical tools for data analysis, including Probability-Density-Function and Continuous Wavelet analysis methods, to provide insights into daytime and nighttime activity structures in different mood states. We observed that in depression and mania, nighttime activity is more structured compared to normal nights. Regarding the days, we can see that depression, normal activity, and mania show increasingly more pronounced levels of structural complexity, in that order. Based on these findings, we performed a Continuous Wavelet analysis for single nights preceding normal, manic and depressive days, respectively, in order to give a quantitative prediction for mood switches. From the structure of the wavelet intensity spectra of the nocturnal activities for the “transition” nights, we could successfully establish the probability of transitions to depressive and manic episodes. Bearing in mind that our results are based on an exceptionally long, but still individual case study, which obviously represents a limitation towards generalizability, we can safely state that the intensity spectra derived from Continuous Wavelet analysis can serve as a quantitative measure of these differences, and suggested to give a solid basis for the prediction of mood-state transitions in Bipolar Disorder. Conclusions Our main findings, based on sensitive statistical tools imply that successful prediction of mood switches following longer or shorter normal episodes in Bipolar Disorder is possible by a proper analysis of nocturnal actigraphy signals. The CWA- based approach outlined here represents a novelty, and expected to have important methodological implications for psychiatric practice.