I am running an accelerometer study on stroke patients to see quality of sleep on their recovery, using light effect and so on.
We have put accelerometer on a group of patients for 48 hours, now we need to analyse the data for the amount of movement during sleep and so on. Can anyone suggest how we do this?
Dear Behdad,
Most of the actual generation of actimeters works with 3 axis. If your focus is to distinguish between sleep and wake activity, it would be prefered to look for a validated algorithm, such as Avi Sadeh. Additionaly, there are many possible types of analysis, from cosinor (or any kind of curve adjustment) or non-parametric tools (IS, IV and amplitude - Van Someren´s protocol). Please, take a look at this selected publication.
Best,
Mario
Article A fresh look at the use of nonparametric analysis in actimetry
Dear Behdad,
With regard to circadian rhythm and sleep-wake analysis, Van Someren has published several important papers (e.g., the 1996, 1999, & 2007 papers using non-parametric analyses in older adults with Alzheimer's). Of particular interest, is the 2007 paper showing not only "day to day" variability, but also "week to week" variability - highlighting the importance of extended recordings. They recommend recordings of more than 7 days, with 14 days being the preferred duration.
I realize you are looking at a different population and this may not be overly helpful to you given you've probably already completed data collection... but thought you might want to know :)
Best regards,
Glenn.
Where is monitor being worn? Wrist? Suggest you contact a guy by name of Vincent van Hees at Newcastle University who has done a lot of work with accelerometry data worn on the wrist for energy expenditure. Cannot give you give email address here but Google will show you the way.
Signal vector magnitude might be the way forward initially (before getting into any complex analysis) - setting thresholds and trying to quantify "activity" that way. Spectrum analysis might also be helpful.
Dear Alan, what type of accelerometer did you use to assess the data? In what format do you have the data? We have developed some software tools to analyze physical activity measured with accelerometers, this includes also some analysis regarding sleep. Depending on the type of data you have it could be possible to use this tools.
Best regards, Stefan
- Badges
- Science topic