Using Spatio-Temporal Data Mining to Study Community Mobility of Individuals with Disabilities

Introduction: Community mobility and participation are important to the health and well-being of individuals with disabilities. Recently, there has been an interest in understanding how Global Positioning Systems (GPS) can be used to monitor mobility and participation of these individuals. Methods: In our study, we attempted to measure community mobility and participation of 5 individuals with psychiatric disabilities over a two-week period, using the mobile application AccuTracking on GPS-enabled cellular phones. Location data were collected at one-minute intervals and sent to AccuTracking’s secure online database. Once data were gathered, we examined the feasibility of using ST-DBSCAN, a relatively new data mining density-based clustering algorithm which enables the detection of spatio-temporal clusters from GPS data in the presence of noise, to calculate a number of variables of interest from the data. Results: After settling on appropriate values of ST-DBSCAN parameters (spatial distance, temporal distance and minimum number of points per cluster), we were able to use ST-DBSCAN output to create the following variables: 1) number of destinations each individual had over the course of the study and on each day; 2) the amount of time each individual spent at each of these destinations; 3) the amount of time each individual spent in transit (i.e., not at a destination); and 4) the total distance each individual traveled over the course of the study. We were able to detect differences on each of these variables across individuals, as well as within-individual variability on different days. Implications: Despite certain limitations, ST-DBSCAN seems to be an appropriate tool for identifying destinations of individuals with disabilities, and to help with the calculation of other variables of interest, including time spent at destinations and in transit, and the total distance traveled.