Sleep and Physical Activity Patterns in Bipolar Disorder Episodes: Are Changes in Mean and Variability Good Predictors?

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F68407700%3A21230%2F25%3A00388133" target="_blank" >RIV/68407700:21230/25:00388133 - isvavai.cz</a>

Výsledek na webu

<a href="https://doi.org/10.1111/bdi.70046" target="_blank" >https://doi.org/10.1111/bdi.70046</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1111/bdi.70046" target="_blank" >10.1111/bdi.70046</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Sleep and Physical Activity Patterns in Bipolar Disorder Episodes: Are Changes in Mean and Variability Good Predictors?

Popis výsledku v původním jazyce

Introduction: Actigraphy provides an objective method for identifying clinical states in individuals with bipolar disorder (BD). Rest-activity-rhythm (RAR) features are known predictors, particularly their mean and variability. This study explores whether time-variability is a predictor itself or just due to multicollinearity with mean values. Method: We analyzed 2-year actigraphy data from 326 BD subjects (235 ± 214 days), including 25 daily features (10 daytime, 15 sleep) related to RAR and fragmentation. To reduce mean-variability dependency, we applied Box-Cox normalization. Clinical states were determined using MADRS, YMRS, and weekly self-assessments, forming mania-remission (34 subjects, 31 ± 21 weeks) and depression-remission (58 subjects, 32 ± 22 weeks) datasets. We computed weekly mean (MEAN) and weekly standard deviation (SD) for each feature in the original and normalized data. A mixed-effects logistic regression model (α = 0.05) with a single predictor and subjects as a random effect assessed MEAN and SD in distinguishing clinical states. Results: In mania (respectively, depression), 4/10 (4/10) daytime and 6/15 (8/15) sleep features both MEAN-based and SD-based models were significant pre-normalization, from which 4/4 (2/4) and 4/6 (6/8) remained significant after normalization. The normalization decreased correlations on average by 0.26 (daytime) and 0.38 (sleep) in mania-remission and by 0.29 and 0.42 in depression-remission. Conclusion: Some actigraphy-based RAR variability predictors are significant due to collinearity with mean values. However, most remain strong standalone predictors of BD episodes, even after normalization. In mania-remission, all daytime-related models remained significant, while some sleep-related models lost significance. In depression-remission, daytime models were more affected. Correlation decreased more in sleep-related features, suggesting greater sensitivity to normalization.

Název v anglickém jazyce

Sleep and Physical Activity Patterns in Bipolar Disorder Episodes: Are Changes in Mean and Variability Good Predictors?

Popis výsledku anglicky

Introduction: Actigraphy provides an objective method for identifying clinical states in individuals with bipolar disorder (BD). Rest-activity-rhythm (RAR) features are known predictors, particularly their mean and variability. This study explores whether time-variability is a predictor itself or just due to multicollinearity with mean values. Method: We analyzed 2-year actigraphy data from 326 BD subjects (235 ± 214 days), including 25 daily features (10 daytime, 15 sleep) related to RAR and fragmentation. To reduce mean-variability dependency, we applied Box-Cox normalization. Clinical states were determined using MADRS, YMRS, and weekly self-assessments, forming mania-remission (34 subjects, 31 ± 21 weeks) and depression-remission (58 subjects, 32 ± 22 weeks) datasets. We computed weekly mean (MEAN) and weekly standard deviation (SD) for each feature in the original and normalized data. A mixed-effects logistic regression model (α = 0.05) with a single predictor and subjects as a random effect assessed MEAN and SD in distinguishing clinical states. Results: In mania (respectively, depression), 4/10 (4/10) daytime and 6/15 (8/15) sleep features both MEAN-based and SD-based models were significant pre-normalization, from which 4/4 (2/4) and 4/6 (6/8) remained significant after normalization. The normalization decreased correlations on average by 0.26 (daytime) and 0.38 (sleep) in mania-remission and by 0.29 and 0.42 in depression-remission. Conclusion: Some actigraphy-based RAR variability predictors are significant due to collinearity with mean values. However, most remain strong standalone predictors of BD episodes, even after normalization. In mania-remission, all daytime-related models remained significant, while some sleep-related models lost significance. In depression-remission, daytime models were more affected. Correlation decreased more in sleep-related features, suggesting greater sensitivity to normalization.

Druh

O - Ostatní výsledky

CEP obor

—

OECD FORD obor

20601 - Medical engineering

Projekt

<a href="/cs/project/EH22_008%2F0004643" target="_blank" >EH22_008/0004643: Dynamika mozku</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů