I have completed a study with older adults in care homes monitoring sleep and light patterns using actigraphy. I have a mixed set of outputs - some with a regular sleep cycle and others more disrupted patterns. I'm trying to best understand sleep quality and the most relevant variables to extract from the software (Philips respironic actiwatch spectrum). I'm not sure if setting a custom interval would help best or if the pattern shown should determine how to look at the stats. Any thoughts of good references welcomed.
You could approach this problem by first remembering that actigraphy does not actually measure sleep – it measures activity. You can stay ‘closer to the data’ of your analytic approach emphasizes the actual movement counts, rather than the processed estimates of what the software calls ‘sleep’. Also, find a way to include the entire 24-hour period of data, rather than only the sleeping period.
The attached study by van Dijk et al., (2012) may help shed some light.
This paper might help. the paper illustrates the method for actigraphy measures in nursing home patients. Some of the published applications are mentioned in the discussion section.
Article The sigmoidally transformed cosine curve: A mathematical mod...
I should add that the other authors, Sonia Ancoli-Israel, Jennifer Martin and Phil Gehrman have used the method in other papers.
Besides the circadian rhythm modeling, we scored the actigraphy methods for sleep/awake using a published algorithm, and looked at the distributions of sleep/wake epochs and looked at those distributions for several periods of the day such as "in bed" and "out of bed" time.
Thanks for your comments and the paper Matthew - I much appreciate your input!
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Computer Software