Our ability to quantify human health has improved dramatically over the years, but a particular area that has historically lagged behind is human activity and behavior. A primary obstacle hindering this endeavor is the ability to capture accurate and objective human behavior data. The recent explosion of wearable sensor technology provides potential solutions to this problem. Smartphones in particular are ubiquitous and contain an array of built-in sensors. Daynamica is a smartphone application that uses machine learning and other algorithms to automatically convert raw sensor data collected by the phone into a daily summary of a user's activity.

Time is partitioned into a sequence of activities, which which take place at fixed locations and capture how people spend the majority of their time, and trips, which capture how we move from activity to activity. Examples of activities detected by Daynamica include home, work, eat out, shopping, and more. Trips are classified by the mode of transportation: car, bus, rail, bike, or walk. Users correct incorrectly inferred data and supplement the data with additional information on positive and negative emotions, companionship, and other trip and activity properties that cannot be inferred directly from sensor data. The figures illustrate the Daynamica data collection process, along with a sample of real data from a participant in a Minneapolis-area study.

My Research

My research focuses on developing new statistical methods to leverage unique features of human activity data, summaries of which can be found at the links below.

In addition to my dissertation work, I have been an active member of the Daynamica research group since 2017, contributing to the ongoing improvement of the Daynamica smartphone app and to its application in multiple studies. Contributions have included:

1. Statistical analysis of Daynamica data

2. Creating and implementing an interactive dashboard combining and visualizing Daynamica and ankle-mounted accelerometer data

3. Developing new data visualizations for exploring unique features of the Daynamica data

4. Creating automated pipelines for processing and cleaning data generated by Daynamica to prepare it for analysis

5. Planning and execution of studies involving Daynamica data collection.