Skip to main content

Wearable Data Analysis Platform

Department​

Math

Summary​

The lab needed data to refine and test their Circadian Rhythms Models, so our group created iPhone and Android apps that allowed users to donate their heart rate and step data and get an analysis back from the lab in return. The smartphone sends data to a server/database, and it is processed by multiple algorithms, some of which generate reports for the users.

Key Benefits to the Lab​

Using mobile applications published to the Apple and Android app stores allowed the lab to collect data from hundreds of users without having to directly recruit study participants The data server’s MATLAB plug-in interface minimized time researchers had to spend integrating their analysis scripts with the databases

Details​

The Wearable Data Analysis Platform (WDAP) is made of three main components: a pair of mobile apps, a server that hosts a database and API server, and a series of servers that process the data. These components work together to create a system for collecting, storing and processing data that the lab can use for their current and future research.

The interface for outside users is a native application for their iPhone or Android phone. These apps host surveys that collect demographics data, and allow users to share either their FitBit or Apple HealthKit data. This data is uploaded anonymously, and allows the lab to send results back to users, which include custom plots based on the lab’s analysis.

The server hosts a database of all submitted data, provides API access for the mobile apps, and displays results for the users.

Once data is deposited to the database, a set of processing servers run the lab’s algorithms on each set of data. This process runs continuously, and constantly checks for new data to pass to the algorithms. We created an easy to use interface for the grad students and post docs in the lab, so that they can provide their code in any programming language (usually MATLAB), and their code reads and writes csv files. Our scripts handle all of the database access so the researchers can focus on their algorithms and not worry about the details of the server, while still accessing and processing the large volume of live data on the server. Working closely with the researchers and the MATLAB code they wrote was one of the key parts of this project, so we made it as easy as possible for them to work with.