Name
Abstract | ||
|---|---|---|
Lower physical activity and sleep problems are linked to depressive and anxiety disorders and can be assessed objectively with wrist-worn actigraphy devices that measure gross motor activity. Standard methods to model 24-h actigraphy are often based on summary measures such as activity counts per day or parametric functional curves (e.g., cosinor functions), losing important information that may be a hallmark of the circadian impairment in psychiatric disorders. Functional Principal Component Analysis (fPCA), a powerful data-driven technique, may better describe daily activity patterns and provide a greater insight into the relationship with depression and anxiety. We assessed patterns of daily activity extracted with fPCA and will evaluate how they are associated with DSM-IV depression and anxiety disorders. Two-week actigraphy data of 367 participants with current (n=94), remitted (n=176) or no (n=90) depression and/or anxiety were obtained from the Netherlands Study of Depression and Anxiety. Daily minute-to-minute activity data were represented with a set of nine Fourier-based functions and summarized with a smaller set of functional principal components with the R package fda. Preliminary analyses resulted in four functional principal components explaining 77.4% of variability (Fig. 1): 1) fPCA1 represented high (+) versus dampened and shifted delayed (-) activity; 2) fPCA2 described morning (+) versus evening (-) activity; 3) fPCA3 showed biphasic (+) versus monophasic (-) activity; 4) fPCA4 represented biphasic morning (+) versus biphasic evening (-) activity. First analyses on the association of the extracted features with depression and anxiety using GEE analysis seem to indicate that current depression and anxiety are significantly associated with lower fPCA1 (?= -0.298, standard error (se) = 0.042, p = 0.006) but not with the other components. Further analyses will be conducted to evaluate if specific depression and anxiety characteristics (e.g. age of onset, duration, comorbidity) are associated with the components. |
Year | DOI | Venue |
|---|---|---|
2018 | 10.1109/eScience.2018.00053 | 2018 IEEE 14th International Conference on e-Science (e-Science) |
Keywords | Field | DocType |
actigraphy,functional analysis,Principal Component Analysis,depression,anxiety | Functional principal component analysis,Data mining,Actigraphy,Computer science,Evening,Anxiety,Comorbidity,Audiology,Morning,Standard error,Gross motor skill | Conference |
ISSN | ISBN | Citations |
2325-372X | 978-1-5386-9157-1 | 0 |
PageRank | References | Authors |
0.34 | 0 | 9 |
Name | Order | Citations | PageRank |
|---|---|---|---|
| Sonia Difrancesco | 1 | 0 | 0.34 |
| Vadim Zipunnikov | 2 | 7 | 1.84 |
| Robert A. Schoevers | 3 | 0 | 0.34 |
| H. Riese | 4 | 0 | 0.68 |
| Niki Antypa | 5 | 0 | 0.34 |
| Brenda W. J. H. Penninx | 6 | 2 | 0.81 |
| Kathleen R. Merikangas | 7 | 0 | 0.34 |
| Albert A. M. van Hemert | 8 | 0 | 0.34 |
| Femke Lamers | 9 | 0 | 0.34 |