Statistical Models for the Time-Varying Effect of Mobile Health Interventions, with Application to HeartSteps

Statistics/Computer Science

This project involves the development of statistical methods to model and analyze the time-varying effect of the mobile health activity suggestions on users’ step counts. A challenge is that the minute-level step count has many zeros ("zero-inflated"). The student would build a test bed that generates simulated zero-inflated data. The test bed will be used for testing the statistical methods being developed. The student will also help develop a statistical method for handling zero-inflated outcome variables in mobile health studies (namely, a generalized estimating equations type method for zero-inflated outcomes). Interested students must have a strong computational and statistical background.