A Bayesian Framework for Individual Exposure Estimation on Uncertain Paths

Abstract

Current map matching and path reconstruction algorithms exhibit high success rates on dense data. We present a framework for estimating path selection probabilities from extremely sparse GPS data for the purpose of estimating a “measurement of interest” that varies with path and travel time. This work is motivated by limitations involved in applications such as environmental exposure modeling for medical patients. Our contributions are two-fold; first we propose a general Bayesian framework for path selection estimation that is applicable at both population and individual levels, and second, we provide extensive experiments on real and synthetic data that demonstrate the accuracy and robustness of the proposed algorithm and model.