The Implementation of a GPS-Based Location-Tracking Smartphone App in South Africa to Improve Engagement in HIV Care: Randomized Controlled Trial

Prevention of adverse HIV treatment outcomes: machine learning to enable proactive support of people at risk of HIV care disengagement in Tanzania

OBJECTIVES

This study aimed to develop a machine learning (ML) model to predict disengagement from HIV care, high viral load or death among people living with HIV (PLHIV) with the goal of enabling proactive support interventions in Tanzania. The algorithm addressed common challenges when applying ML to electronic medical record (EMR) data: (1) imbalanced outcome distribution; (2) heterogeneity across multisite EMR data and (3) evolving virological suppression thresholds.

DESIGN

Observational study using a national EMR database.

SETTING

Conducted in two regions in Tanzania, using data from the National HIV Care database.

PARTICIPANTS

The study included over 6 million HIV care visit records from 295 961 PLHIV in two regions in Tanzania's National HIV Care database from January 2015 to May 2023.

RESULTS

Our ML model effectively identified PLHIV at increased risk of adverse outcomes. Key predictors included past disengagement from care, antiretroviral therapy (ART) status (which tracks a patient's engagement with ART across visits), age and time on ART. The downsampling approach we implemented effectively managed imbalanced data to reduce prediction bias. Site-specific algorithms performed better compared with a universal approach, highlighting the importance of tailoring ML models to local contexts. A sensitivity analysis confirmed the model's robustness to changes in viral load suppression thresholds.

CONCLUSIONS

ML models leveraging large-scale databases of patient data offer significant potential to identify PLHIV for interventions to enhance engagement in HIV care in resource-limited settings. Tailoring algorithms to local contexts and flexibility towards evolving clinical guidelines are essential for maximising their impact.

Peripartum mobility and maternal/child separation among women living with HIV in South Africa

This prospective cohort study investigated the mobility patterns of 200 pregnant and postpartum women living with HIV in South Africa. Participants were enrolled during their third trimester from routine antenatal care near Cape Town, South Africa, and followed for six months postpartum. Quantitative data were collected at enrollment and follow-up. Mobility (self-reported) was common among the participants, despite the brief study period and the concurrent COVID-19 pandemic. While most reported stability in their current residence, 71% had a second main residence, primarily in the Eastern Cape (EC). Participants had a median of two lifetime moves, motivated by work, education, and family life. During the study period, 20% of participants met the study definition of travel (>7 days and >50 km), with trips predominantly to the EC, lasting a median duration of 30 days. Over one-third of participants with other living children reported that these children lived apart from them, with the mother's family being primary caregivers. These findings emphasize the need for targeted interventions to support continuity of care for mobile populations, particularly peripartum women living with HIV. The study contributes valuable insights into mobility dynamics and highlights unique barriers faced by this population, contributing to improved HIV care in resource-limited settings.

Leveraging Ecological Momentary Assessment Data to Characterize Individual Mobility: Exploratory Pilot Study in Rural Uganda

BACKGROUND

The geographical environments within which individuals conduct their daily activities may influence health behaviors, yet little is known about individual-level geographic mobility and specific, linked behaviors in rural low- and middle-income settings.

OBJECTIVE

Nested in a 3-month ecological momentary assessment intervention pilot trial, this study aims to leverage mobile health app user GPS data to examine activity space through individual spatial mobility and locations of reported health behaviors in relation to their homes.

METHODS

Pilot trial participants were recruited from the Rakai Community Cohort Study-an ongoing population-based cohort study in rural south-central Uganda. Participants used a smartphone app that logged their GPS coordinates every 1-2 hours for approximately 90 days. They also reported specific health behaviors (alcohol use, cigarette smoking, and having condomless sex with a non-long-term partner) via the app that were both location and time stamped. In this substudy, we characterized participant mobility using 3 measures: average distance (kilometers) traveled per week, number of unique locations visited (deduplicated points within 25 m of one another), and the percentage of GPS points recorded away from home. The latter measure was calculated using home buffer regions of 100 m, 400 m, and 800 m. We also evaluated the number of unique locations visited for each specific health behavior, and whether those locations were within or outside the home buffer regions. Sociodemographic information, mobility measures, and locations of health behaviors were summarized across the sample using descriptive statistics.

RESULTS

Of the 46 participants with complete GPS data, 24 (52%) participants were men, 30 (65%) participants were younger than 35 years, and 33 (72%) participants were in the top 2 socioeconomic status quartiles. On median, participants traveled 303 (IQR 152-585) km per week. Over the study period, participants on median recorded 1292 (IQR 963-2137) GPS points-76% (IQR 58%-86%) of which were outside their 400-m home buffer regions. Of the participants reporting drinking alcohol, cigarette smoking, and engaging in condomless sex, respectively, 19 (83%), 8 (89%), and 12 (86%) reported that behavior at least once outside their 400-m home neighborhood and across a median of 3.0 (IQR 1.5-5.5), 3.0 (IQR 1.0-3.0), and 3.5 (IQR 1.0-7.0) unique locations, respectively.

CONCLUSIONS

Among residents in rural Uganda, an ecological momentary assessment app successfully captured high mobility and health-related behaviors across multiple locations. Our findings suggest that future mobile health interventions in similar settings can benefit from integrating spatial data collection using the GPS technology in mobile phones. Leveraging such individual-level GPS data can inform place-based strategies within these interventions for promoting healthy behavior change.

Informing Acceptability and Feasibility of Digital Phenotyping for Personalized HIV Prevention among Marginalized Populations Presenting to the Emergency Department

For marginalized populations with ongoing HIV epidemics, alternative methods are needed for understanding the complexities of HIV risk and delivering prevention interventions. Due to lack of engagement in ambulatory care, such groups have high utilization of drop-in care. Therefore, emergency departments represent a location with those at highest risk for HIV and in highest need of novel prevention methods. Digital phenotyping via data collected from smartphones and other wearable sensors could provide the innovative vehicle for examining complex HIV risk and assist in delivering personalized prevention interventions. However, there is paucity in exploring if such methods are an option. This study aimed to fill this gap via a cross-sectional psychosocial assessment with a sample of N=85 emergency department patients with HIV risk. Findings demonstrate that although potentially feasible, acceptability of digital phenotyping is questionable. Technology-assisted HIV prevention needs to be designed with the target community and address key ethical considerations.

Exploring mobility data for enhancing HIV care engagement in Black/African American and Hispanic/Latinx individuals: a longitudinal observational study protocol

INTRODUCTION

Increasing engagement in HIV care among people living with HIV, especially those from Black/African American and Hispanic/Latinx communities, is an urgent need. Mobility data that measure individuals' movements over time in combination with sociostructural data (eg, crime, census) can potentially identify barriers and facilitators to HIV care engagement and can enhance public health surveillance and inform interventions.

METHODS AND ANALYSIS

The proposed work is a longitudinal observational cohort study aiming to enrol 400 Black/African American and Hispanic/Latinx individuals living with HIV in areas of the USA with high prevalence rates of HIV. Each participant will be asked to share at least 14 consecutive days of mobility data per month through the study app for 1 year and complete surveys at five time points (baseline, 3, 6, 9 and 12 months). The study app will collect Global Positioning System (GPS) data. These GPS data will be merged with other data sets containing information related to HIV care facilities, other healthcare, business and service locations, and sociostructural data. Machine learning and deep learning models will be used for data analysis to identify contextual predictors of HIV care engagement. The study includes interviews with stakeholders to evaluate the implementation and ethical concerns of using mobility data to increase engagement in HIV care. We seek to study the relationship between mobility patterns and HIV care engagement.

ETHICS AND DISSEMINATION

Ethical approval has been obtained from the Institutional Review Board of the University of California, Irvine (#20205923). Collected data will be deidentified and securely stored. Dissemination of findings will be done through presentations, posters and research papers while collaborating with other research teams.

A Mobile Health Application Using Geolocation for Behavioral Activity Tracking

The increasing popularity of mHealth presents an opportunity for collecting rich datasets using mobile phone applications (apps). Our health-monitoring mobile application uses motion detection to track an individual's physical activity and location. The data collected are used to improve health outcomes, such as reducing the risk of chronic diseases and promoting healthier lifestyles through analyzing physical activity patterns. Using smartphone motion detection sensors and GPS receivers, we implemented an energy-efficient tracking algorithm that captures user locations whenever they are in motion. To ensure security and efficiency in data collection and storage, encryption algorithms are used with serverless and scalable cloud storage design. The database schema is designed around Mobile Advertising ID (MAID) as a unique identifier for each device, allowing for accurate tracking and high data quality. Our application uses Google's Activity Recognition Application Programming Interface (API) on Android OS or geofencing and motion sensors on iOS to track most smartphones available. In addition, our app leverages blockchain and traditional payments to streamline the compensations and has an intuitive user interface to encourage participation in research. The mobile tracking app was tested for 20 days on an iPhone 14 Pro Max, finding that it accurately captured location during movement and promptly resumed tracking after inactivity periods, while consuming a low percentage of battery life while running in the background.