JMIR Publications

ABSTRACT

Background:

Smartphones offer a convenient and precise method for studying human mobility at an unprecedented scale, allowing researchers to explore the links between mobility and HIV risk and poor treatment outcomes. However, leveraging smartphone technology to study HIV risk in rural settings presents unique challenges and opportunities.

Objective:

This study evaluates the feasibility of using smartphone GPS technology to record mobility among young adults in rural KwaZulu Natal, South Africa. We also present key lessons learned during the study.

Methods:

The study involved young people aged 20-30 years old and was conducted in two phases from June 2021 to May 2023. In phase I, participants were provided with study smartphones installed with Avicenna research software, a commercially available Android app customized for our study; participants could alternatively install the app on their personal smartphones. Phase II required participants to install the app on their personal smartphones (no longer provided study smartphones). The app used Android’s location services to record location every 5 minutes, later adjusted to every 30-minutes. Location data were uploaded hourly to a secure study server via internet connection, and participants were followed up for six months (26 weeks). In cases where no location data was transmitted for 48-72 hours, participants were contacted via phone or home visits for troubleshooting.

Results:

A total of 207 participants (163 in Phase I) consented to the study and 204 successfully recorded and uploaded location data, yielding over 27 million location points. The number of location points ranged from 53 – 1,351,003 with a median number of 74,865 [IQR 28,471 – 186,578] per participant. The mean location points per week was 28.4% (std. 18.4) out of the expected 336 half hour intervals. In Phase I, percentage of weekly location data peaked at 45.4% in Week 2 but declined by over half to 21.5% by week 26. In Phase II, there were several peaks with more consistent location data towards the latter half of the study, attributed to increased user engagement with the app. Challenges in data collection included interruptions due to poor internet connectivity, frequent app terminations, phone-related issues such as screen malfunctions or lost/broken devices. Solutions implemented included a reverse billing system to address internet access and a "Wheel of Fortune"- lottery game in Phase II to increase app engagement and reduce app termination instances.

Conclusions:

This study demonstrates the feasibility and acceptability of using both study-provided and personal smartphones to collect mobility data in rural settings. Ongoing and frequent troubleshooting efforts helped identify challenges and inform effective solutions, such as the reverse billing system and engagement activities. These insights provide valuable lessons for future research on HIV risk and mobility patterns using smartphone-based technology.