Background: Our understanding of the effects of human movement on dengue virus spread remains limited in part due to the lack of precise tools to monitor the time-dependent location of individuals. We determined the utility of a new, commercially available, GPS data-logger for long-term tracking of human movements in Iquitos, Peru. We conducted a series of evaluations focused on GPS device attributes key to reliable use and accuracy. GPS observations from two participants were later compared with semi-structured interview data to assess the usefulness of GPS technology to track individual mobility patterns.
Results: Positional point and line accuracy were 4.4 and 10.3 m, respectively. GPS wearing mode increased spatial point error by 6.9 m. Units were worn on a neck-strap by a carpenter and a moto-taxi driver for 14-16 days. The application of a clustering algorithm (I-cluster) to the raw GPS positional data allowed the identification of locations visited by each participant together with the frequency and duration of each visit. The carpenter moved less and spent more time in more fixed locations than the moto-taxi driver, who visited more locations for a shorter period of time. GPS and participants' interviews concordantly identified 6 common locations, whereas GPS alone identified 4 locations and participants alone identified 10 locations. Most (80%) of the locations identified by participants alone were places reported as visited for less than 30 minutes.
Conclusion: The present study demonstrates the feasibility of a novel, commercially available GPS data-logger for long-term tracking of humans and shows the potential of these units to quantify mobility patterns in relationship with dengue virus transmission risk in a tropical urban environment. Cost, battery life, size, programmability and ease of wear are unprecedented from previously tested units, proving the usefulness of GPS-dataloggers for linking movement of individuals and transmission risk of dengue virus and other infectious agents, particularly in resource-poor settings.