Sleep measuring apps?

[Viktor Kaldo]

Dear app-experts out there!

I have been doing a lot of ICBT research on Insomnia. So far, we have used actigraphs to get a more objective measure of sleep pattern. Traditionally, besides handling the logistics with sending actigraphs back and forth, this includes handling a lot of data and manual aspects of the scoring.
So far, I have been a bit suspicious about the reliability of mobile app solutions for this, regardless if they use the built-in accelerometer in the phone or a Bluetooth connected external device (for example a Fitbit on the arm). Does anyone now if the accuracy of the available apps nowadays are good enough to use for scientific purposes?

It would also be interesting to know if someone have tested an app for this purpose, or as a part of a CBT for insomnia intervention? When used as a therapeutic tool, the accuracy might not be as crucial, but then the user interface and type of feedback is even more important.

All the best,
Viktor Kaldo

[Pepijn Van De Ven]

Hi Viktor,

Interesting question! Instead of an answer, I’d like to give you my musings. I am sure you are well aware of the various validation studies out there that have compared consumer activity trackers to established means of sleep monitoring (actigraphy and polysomnography). From what I have seen, generally the conclusion is that there is reasonable correlation between validated devices and consumer activity trackers, but that there tends to be under or overestimation of various sleep parameters/times resulting in conclusions that consumer devices ‘may’ be of use. Smart phone apps (see e.g. https://www.ncbi.nlm.nih.gov/pubmed/26446248) are likely inferior when taken into consideration in these comparisons.

However, consumer activity trackers also provide certain benefits. As you also mention, data collection can be an issue with more traditional devices and I am sure that often the size of a study will be dictated by the number of devices available. For this reason, perhaps consumer devices are an interesting alternative. They are cheaper, many people already have one, in some cases allow data to be pulled down from the cloud with very little extra (development) effort, and making a (web or mobile) user interface for your users is often reasonably straightforward.

We tend to compare new technologies with the gold standard in the area of interest, and I think that in most cases the consumer devices will then be inferior to some extent. However, perhaps we don’t always need the gold standard (as you also suggest Viktor) and silver may be sufficient. So perhaps we should turn things around and look at what type of accuracy is required to reach a certain goal and whether this allows us to look at cheaper alternatives to validated actigraphy?

All the best,

Pepijn

[Author]

Posts