ABSTRACT
Much insight on environment-health nexus can be gained from mining GPS trajectory data. Despite much interest in understanding the environmental context of physical activity using GPS and accelerometers among health researchers, lack of consistent methods for processing GPS data inhibits further progress. A fully automated approach was developed to process uncertain GPS trajectory data, and integrate the processed GPS data with activity data. The approach filled missing location data (or time gaps) in GPS trajectory before GPS data was segmented into episodes by movement state. The approach was developed utilizing data from 39 study participants who each agreed to carry a GPS data logger and activity monitor for two consecutive days. Experimentation results show that the proposed method enabled 99.7% of GPS data to be matched with activity data and GPS trajectory was partitioned into episodes of stops and trips with 95.84% accuracy. Categorized location of physical activity (such as home, work and transportation) was also calculated from the synchronized GPS/activity data. This chapter highlights the importance of data quality in GPS trajectory computing. It also demonstrates how big data analytics can be harnessed to improve exposure assessment and to inform policies promoting health.
