Activity tracking devices such as the FitBit, Jawbone, and Microsoft Band are rapidly rising in popularity as the Internet of Things trend grows. Many of these devices claim to have life-changing health benefits. This study examines the accuracy of these devices and apps by comparing each device's sleep-stage data and accelerometer values with each other and to a research-standard actigraphy watch. This project is part of a larger project on personal informatics at the Brown HCI department.
My first task is to use all devices simultaneously while sleeping to gather data -- that means wearing six activity tracking wristbands and turning on four different apps at night, for two weeks straight! This is definitely the silliest part, and it's always fun to talk about. :-)
The next phase of the project is to extract data from each of these devices, and is possibly the most difficult. Many of these devices do not make minute-by-minute sleep data (e.g. sleep vs. awake, deep sleep vs. light sleep) readily or easily available. This part has so far involved lots of API use as well as a little bit of hacking here and there. I'll also be formatting all this data into one uniform pattern in order to process it in the next part.
Next is visualizing the data to be easily understood by both researchers and consumers alike, since the data itself (as shown above) is pretty hard to read otherwise. This will involve lots of d3.js, lots of data cleaning, and lots of research. I'm pretty excited for this part since I've been itching to get interesting data to create awesome visualizations. I'm also going to have documentation so that the HCI group can use this data in future papers and studies, and maybe making that readily available for other researchers to investigate!
Some slides for my mid-semester presentation on this project can be found below.