Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 May 13.
Published before final editing as: Int J Geogr Inf Sci. 2026 Mar 17:10.1080/13658816.2026.2641743. doi: 10.1080/13658816.2026.2641743

Pharos: a mobile GIS application for network assessment and ground-truth mapping to support mHealth study design in low-resource settings

Carson P Moore a, Natalie N Robbins b, Gladys Odhiambo c, Kennedy Andiego c, Meredith Odhiambo c, Fredrick Rawago c, Rosemary Musuva c, Maurice Odiere c, David Wright a, Thomas Scherr a
PMCID: PMC13166115  NIHMSID: NIHMS2161662  PMID: 42131713

Abstract

Mobile health (mHealth) is a promising tool for improving healthcare access, particularly in low-resource settings. However, limited mobile accessibility and poor connectivity remain significant barriers to implementing mHealth interventions in these regions. To address these challenges and support the development and scalability of mHealth studies, we developed Pharos, a mobile GIS application designed to assess network coverage and facilitate ground-truth mapping. Pharos autonomously measures spatiotemporal variations in mobile network signal strength and enables precise mapping of critical landmarks and environmental features. We deployed Pharos in a four-county region along the shores of Lake Victoria in Western Kenya as part of a preparatory phase for a large-scale mHealth study focused on schistosomiasis control. Over six months and 10,000 km2, Pharos collected high- resolution data on network performance and landmark locations, generating a comprehensive dataset that links network availability with environmental features. These results provide essential insights for planning and implementing mHealth interventions in low-resource settings, with potential applications in infectious disease surveillance and other global health initiatives.

Keywords: Mobile health, mobile mapping, environmental epidemiology

Introduction

Since its inception in the 1970s, mobile networking has presented a unique opportunity to increase connection among individuals worldwide, particularly those living in low- and middle-income countries (LMICs) (McCool et al. 2022). Readily available integration of global positioning systems (GPS) in our phones coupled with the expansion of mobile networking allows for real time data collection and monitoring in areas where formal, ubiquitous geospatial data may be lacking and where traditional mapping efforts may be time and labor intensive (Siedner et al. 2012; Nowak et al. 2020; Nyapwere et al. 2021). In 2015, the United Nations presented their 17 Sustainable Development Goals (SDGs), and the mobile networking industry was among the first to commit to achieving the SDGs (Thales 2023). In 2023, nearly 10 years after the inception of the SDGs, there were an estimated 5.4 billion people (67% of the global population) connected to the internet, and 58% of all internet traffic came from a mobile device (Statista 2022; International Telecommunication Union 2023).

Despite stumbling blocks in creating sustainable progress towards global equity, due in part to the COVID-19 pandemic, global conflict, increasing climate disasters, and cost-of-living challenges, mobile networking has boomed in LMICs (GSMA 2023a, 2023b; Jones et al. 2017). For individuals living in LMICs, mobile connectivity is the primary, and often only, method to access the internet. In 2022, of the countries in which mobile connection was the largest portion of all web services, 8 of the top 10 were classified as LMICs (Statista 2022; Wellcome 2022). In Kenya, the country with the 8th highest rate of mobile-based internet traffic globally, 99% of the population has mobile broadband coverage, and 81% are actively connected to mobile broadband (Statista 2022; GSMA 2023a, 2023b). Further, the ubiquity of mobile connectivity has proven a critical tool in several fields, including geographic information systems (GIS), mobile health (mHealth), and more.

Amid this rise in mobile connectivity, mHealth tools have already become essential in LMICs for expanding healthcare access, strengthening data collection, and boosting public awareness of health services and issues (Lewis et al. 2012; Osei and Mashamba-Thompson 2021; McCool et al. 2022). mHealth is defined as any medical or public health practice that is supported by or performed with a mobile device (including but not limited to mobile phones, tablets, and personal digital assistants (PDAs)) (Lewis et al. 2012). However, the most common devices utilized by mHealth programs are mobile phones (71%), compared to computers (39%), cameras (7%) and tablets (6%) (Lewis et al. 2012). While many of the most common mHealth interventions utilize short message service (SMS) to achieve program goals due to the low cost and simple deployment, in recent years, the number of applications and software packages for mHealth has expanded considerably (Dêglise et al. 2012; Free et al. 2013; Orr and King 2015). In 2015, the number of mHealth apps available on the Apple App Store and Google Play Store exceeded 100,000, and that number has experienced substantial growth since the start of the COVID-19 pandemic (Sama et al. 2014; Xu and Liu 2015; Pires et al. 2020).

The current proliferation of mHealth apps in resource-limited and rural areas of the world offer a unique opportunity for both passive and active collection of GPS information that can geo-enrich studies, provide important spatial context for other metrics being collected, and allow post-data collection spatial statistics to be calculated to understand the feasibility of studies. This type of work, combining mHealth and GIS, has begun to be utilized in rural areas of developed countries, but has yet to be investigated in detail in more resource-constrained, infectious disease-endemic settings (Rogers and Randolph 2003; Kamel Boulos and Koh 2021; Burger et al. 2024).

Schistosomiasis, one such infectious disease that is endemic to many tropical and subtropical regions, is a strong candidate for GIS- and mHealth-based interventions due to its unique transmission cycle and dependence on environmental conditions. Schistosomiasis is an infectious parasitic disease of poverty affecting nearly 240 million people worldwide, with an estimated 1.64 million disability-adjusted life years (DALYs) lost in 2019 (Gryseels et al. 2006; Colley et al. 2014; Centers for Disease Control and Prevention 2019). Although fatalities are rare, schistosomiasis can lead to severe symptoms, including anemia, cognitive impairment, infertility, liver fibrosis, and bladder cancer, all of which significantly impact affected individuals and communities (Colley et al. 2014; McManus et al. 2018).

The disease is most prevalent in tropical and subtropical regions, especially in SubSaharan Africa, where inadequate sanitation and limited access to clean water facilitate transmission (Global Burden of Disease, 2020; World Health Organization, n.d.; Vos et al. 2020). Infection occurs through skin contact with freshwater containing blood flukes of the genus Schistosoma, which penetrate the skin upon contact (Pearce and MacDonald 2002; Colley and Secor 2014; Mari et al. 2017; Nelwan 2019). Transmission relies on the presence of intermediate host snails in water bodies—Bulinus for S. haematobium and Biomphalaria for S. mansoni—each of which has unique ecological requirements (Mandahl-Barth 1957; Woolhouse and Chandiwana 1989; Friani et al. 2023).

For environmentally-focused diseases like schistosomiasis, there is strong potential for GIS-assisted mHealth programs (Wood et al. 2019; Blanford and Jolly 2021). GIS can help identify high-risk areas by mapping infection hotspots, environmental risk factors, and patterns of population movement, which can guide targeted treatment campaigns and water sanitation initiatives. Combining mHealth with GIS offers a promising approach for managing schistosomiasis in low-resource settings, where spatial information is crucial for effective public health strategies.

However, despite the well-documented parallel uptake of mobile networking and mobile health applications, some stakeholders still maintain a level of skepticism about the utility of mobile-connected research in these areas in the field (Mechael et al. 2010; Istepanian and AlAnzi 2020). This dissatisfaction primarily stems from a failure of many existing mHealth programs to scale up to broad-scale interventions, a phenomenon which has been termed ‘pilotitis’ (Lemaire, 2011). Pilot testing is a process in which small-scale testing is undertaken to determine the feasibility of deployment at a large scale. This process allows researchers to assess the procedures, principles, strategies, and design of their interventions before committing significant capital and time to a specific broad-scale method. As a representative example, 36 mobile-connected pilots were undertaken in Uganda between 2008 and 2009, and only 13 (36%) managed to progress beyond the initial pilot stage (Greve et al. 2022).

To address these gaps between pilot testing and large-scale implementation, we developed Pharos, a native application designed to proactively map mobile network performance in potential study sites by collecting both spatial and temporal mobile network data alongside critical local environmental features. While some mHealth applications can leverage asynchronous data transmission and storage with opportunistic sync strategies (e.g., Apache CouchDB, PouchDB, AWS DataStore) to tolerate degraded networks, applications that utilize SMS or more compute-intensive processing (i.e., medical image analysis, machine learning) depend on server connectivity for full functionality. Additionally, certain non-portable hardware may offer limited utility without consistent network access. As such, formative studies that assess network performance can provide valuable insights to improve the design and deployment of mHealth interventions, ultimately supporting both scale-up and long-term success (Boonchieng et al. 2021).

In addition to understanding spatiotemporal characteristics of local mobile networks, the Pharos application also allows researchers to collect granular geographic information related to the use-case of their studies. In this work, we collected geospatial data relevant to population studies (e.g., the locations of population centers such as schools, churches and market centers) as well as disease-related landmarks important for understanding schistosomiasis transmission (e.g., water bodies, snail colonies, and plant life). Further, we were able to map the locations of cell towers in the area and compare this data to the passively collected recurring network data, and compare Pharos-collected GIS data to current gold-standard open-access GIS data. To date, there are few—if any—publicly-available datasets that incorporate these features. The ability to collect and distribute this data will be critical for understanding the spread and risk of disease in the study regions, further strengthening the case for GIS-informed mHealth interventions in the future.

Methods

Hardware

The application was designed for compatibility with both Android and iOS devices. Although Apple products are less common globally than in the United States, to reduce inter-device variability in the study, Apple iPhone SE (3rd generation, 2022) devices running iOS15 were used for this study (StatCounter Global Stats, 2024). Data collection was performed using each of the three primary cellular networks operating in and around Kisumu, Siaya, Migori, and Homa Bay counties in Western Kenya. To reduce bias and blind results during data dissemination, each phone was assigned a color (red, orange, and blue) and labeled (A, B, C) for the duration of the study.

Software

The Pharos mobile application (app) was written in JavaScript using the React Native framework and deployed through TestFlight, the Apple App Store’s beta testing program. The app has several new screens and features in comparison to a previously utilized application (Beacon) (Scherr et al. 2020) (Figure 1). An overview of the software architecture is shown in Figure 2.

Figure 1.

Figure 1.

Open in a new tab

Application user-interface for the Pharos application. Initially, users are presented with the login screen (A), which directs to the current data home screen (B). On this screen, the most recently collected data point is displayed, as well as an aggregate count of the user’s collected data points for the session. Users can press the blue floating action button at the bottom right of the screen to open the landmark data collection menu (inset, E-F). Users can then navigate to the aggregate data screen (C) to view a map of collected data, upload collected data to the database, and view the 100 most recent RNMs and landmarks collected. On the settings screen (D), users can tailor their Pharos experience to better match the intended mHealth use case.

Figure 2.

Figure 2.

Open in a new tab

(A) A software architecture diagram for Pharos. (B) A locational map highlighting the four counties in Western Kenya where our study took place.

Upon opening the app, users are prompted to log in using usernames and passwords. They are then directed to a landing screen, which shows the results from the most recently collected data point, as well as a summary of the total number of recurring network measurements (RNMs) and landmarks collected (Figure 1B). RNMs are collected by sending a data packet—in this instance, an image of a rapid diagnostic test (of specific relevance to our planned infectious disease surveillance study)—to be uploaded to a cloud-based Amazon Web Services (AWS) Simple Storage Service (S3) bucket. A different image of the same size is downloaded from S3.

The duration of each of these file transfers is measured within the application and tagged with a GPS location and time stamp. From the ‘Current Data’ landing screen, users can navigate to a ‘View Data’ screen which shows a small summary map, as well as the last 100 RNMs and last 100 landmarks collected (Figure 1C). From this screen, users can upload any collected data to an AWS Aurora relational database by pressing the ‘Upload to Database’ button. In the absence of network connectivity, data persists locally on the user’s device until network connectivity is restored.

Despite the presence of mobile networking in rural areas, it is still likely that some dead zones- areas where no connection is available- will be encountered during mapping. To avoid data loss the user receives an alert informing them that their network connectivity is not strong enough to upload, and to retry the upload when reconnected. Users can also navigate to the ‘Settings’ screen, which provides control over specific operational parameters within the Pharos app (Figure 1D). These settings allow users to adjust the frequency of data collection, file size for uploads and downloads, and the AWS region of the receiving server, which can help simulate different data demands and geographic requirements for future studies. Unlike the previous version of Pharos, which transmitted a fixed data packet to a static server at Vanderbilt University (Nashville, TN, USA), the current Pharos app supports customization options to better approximate various study workloads. For this study, we used a 2.5 MB image of a rapid diagnostic test for uploads and downloads, and the receiving server was set to Amazon’s US-East-1 region (Virginia, USA).

Landmark collection

A notable addition to the Pharos app compared to previous work is the addition of ground-truthing capabilities and geospatial data collection through photography, accessible through a floating action button in the lower right-hand corner of the landing screen (Figure 1E, F). When pressed, a floating menu appears that allows users to specify the type of landmark GPS point they are collecting (e.g., a general cultural landmark, cell tower, or disease-specific landmark). Once the landmark subtype is selected, users are then prompted to enter more information about the landmark, for instance, if they are recording a cultural landmark, they can identify it as a school, church, community health center, market center, or other landmark, and enter a freeform text description to support the entry. Then, the user can capture a photograph of the landmark of interest. Once a landmark has been logged, the user is directed back to the ‘Current Data’ landing screen. Upon data uploads, landmark photographs are stored in a dedicated AWS S3 bucket and photograph meta-data is stored in an AWS Aurora database.

Data collection interface and controls

Recurring network measurement (RNM) data collection is automated and occurs at regular intervals without requiring manual input, provided that RNM collection is toggled on. Users can control this feature in the Settings menu by adjusting the frequency of data collection or disabling it entirely. When enabled, the app autonomously gathers RNM data and associated metadata. For LM measurements, the Pharos app currently supports two main types of data entry: free text and controlled lists via single-select drop-down menus. For validation, landmark subtype options are dynamic, determined from the parent landmark type selected. Landmark classifications within the app are predefined using controlled vocabularies, and, where possible, these classifications align with OpenStreetMap (OSM) terminologies, helping maintain consistency with established GIS data standards. Free text allows users to input additional notes for later review.

Metadata is collected to enhance the usability of both landmark and RNM data. For landmarks, metadata includes GPS coordinates, a timestamp, and an optional free text description. For RNMs, metadata consists of signal strength, upload and download durations, GPS coordinates, and timestamp. Additionally, for each datapoint, information about the cellular device and the network is collected as well: device manufacturer, device name, operating system, operating system version number, cellular carrier name, cellular network generation, and cellular service provider country code. Upon data uploads, landmark photographs are stored in a dedicated AWS S3 bucket, and metadata for both landmarks and RNMs is managed through an AWS Aurora database. This infrastructure supports flexible data collection and allows for future enhancements that may extend the app’s functionality to support a more comprehensive user experience.

The automated RNM data collection in Pharos reduces the burden on field workers by providing a continuous stream of network performance data without requiring manual input, allowing users to focus on other tasks. The landmark input interface is designed to support efficient, flexible data collection in the field, with a user-friendly layout that minimizes input complexity and maximizes data accuracy. Dynamic sub-types further reduce cognitive load by displaying only relevant options based on the selected landmark type. This context-aware, low-burden design approach aligns with human-computer interaction research on mobile data collection, supporting consistent data capture in varying field conditions (Ryan et al. 1999; Pascoe et al. 2000; Alsos 2008). By integrating these design principles, Pharos enhances usability, enabling field workers to efficiently and accurately collect data, even in challenging environments.

Four-county pilot testing

The Pharos application was used to map the network performance and proactively collect geospatial points of interest across four counties in Western Kenya (Table 1) slated for a larger planned mHealth study. In April 2023, three members of the study team (referred to hereafter as the “field team” to differentiate from the “software development team”) were trained in the use of the app and asked to collect data with the target of 30,000 RNM data points. The field team traveled to each of the study counties, collecting RNMs and landmark data for all public primary schools (a focal point of schistosomiasis interventions) in the catchment area of the planned study.

Table 1.

Descriptive population characteristics for each of the four study counties. Adapted from (Kenya National Bureau of Statistics 2019).

County Area (km2) Total population Average household size* Population (age 3+) attending primary school
Siaya  2530 993,183 3.9 261,013
Kisumu  2085 1,155,574 3.8 277,804
Homa Bay  3153 1,131,950 4.3 300,787
Migori  2613 1,116,436 4.6 301,799
Open in a new tab
*

Population and household statistics exclude ‘special populations’ (travelers, outdoor sleepers, individuals that reside in hotels, lodges, or institutions).

Data analysis

Data was downloaded from the AWS Aurora database to dedicated machines for analysis. Analysis was performed in custom Python (v. 3.9.16) code, using the Pandas, Seaborn, Numpy, and Matplotlib libraries. Spatial analysis and visualizations were made using ArcGIS Pro (v3.2.1).

Spatial analysis

To quantify app performance and understand network distribution and strength, cell phone tower data collected in Pharos (n = 121 towers) was used to generate Voronoi diagrams in ArcGIS Pro to represent cell phone tower catchment regions (Konstan et al. 2011; Oliver et al. 2015; Cumbane and Gidôfalvi 2021). A harmonic mean for upload/download speed (mbs/sec) was calculated for each RNM data point, and RNM data was aggregated to the catchment regions and an average harmonic speed was calculated per catchment zone for each cell phone carrier. RNM data with unsuccessful upload and/or download was also stratified by carrier and represented on the map.

To better understand the correlation between snail presence, water presence, and vegetation presence, colocation analysis was run on LM data collected on snails, water, and vegetation using ArcGIS Pro. A K Nearest Neighbors (where K = 8) neighborhood type was run on each permutation.

Results

Mobile Carrier efficacy in Western Kenya

The Pharos application was used to examine the mobile broadband performance using the three primary cell carriers in the study area. A total of 32,500 unique RNMs were collected: 20,319 (62.5%) using Carrier A, 7,457 (22.9%) using Carrier B, and 4,724 (14.5%) using Carrier C (Figure 3A).

Figure 3.

Figure 3.

Open in a new tab

(A) Counts of successful and failed uploads and downloads for each study carrier. (B) Violin plots showing upload and download duration for each study carrier.

Among these carriers, Carrier C was observed to have the highest failure rate (68.5% failed uploads, 63.6% failed downloads) as well as the highest RNM durations (Figure 3B). With outliers (z Score ≥ 3) removed, Carrier A was observed to have the shortest average duration of upload (8.57 s±9) and download (5.28 s±5) events compared to Carriers B (19.0 s ± 13 and 8.80 s ± 9) and C (23.4 s ± 16 and 12.5 s ± 10), with no significant difference between the mean duration for either RNM type for each carrier. Further, the maximum upload and download durations for each of the three carriers were similar (Carrier A: 70.1 and 50.6 s, Carrier B: 70.1 and 50.5 s, Carrier C: 68.9 and 50.6 s) while there was some stratification of the minimum durations across the three carriers (Carrier A: 1.00 and 1.60 s, Carrier B: 2.05 and 2.92 s, Carrier C: 3.06 and 2.94 s). Similar to our previous work in Zambia, where measurements followed informal reports from residents, these findings align with anecdotal reports provided by residents of the region (Scherr et al. 2020).

In addition, spatiotemporal efficacy of the mobile networks in the region was examined. Spatial efficacy was analyzed as a function of cell tower coverage area. There was no significant correlation observed between cell tower presence (n = 121) and strength of network across any of the three studied carriers within the study area (Figure 4).

Figure 4.

Figure 4.

Open in a new tab

Voroni diagrams showing the mean harmonic speed (MBs/second) for each of the three carriers in relation to Pharos-mapped cell towers (n= 121). Upload, download, and both upload/download failures are shown on the map for each carrier.

Temporal efficacy of each carrier was examined for each of the four study counties. During transit to and from study locations, incidental data points were collected out-side of the primary study areas, in neighboring Kakamega, Kericho, Kisii, and Vihiga counties (Figure 5A). However, the volume of data collected from these counties was much lower compared to the study area, and the lack of sufficient data across different timepoints in these locations prevented the creation of meaningful visualizations. Figure 5B and Supplemental Figure 2 show the temporal data by hour of the day collected using each carrier within each of the four study counties.

Figure 5.

Figure 5.

Open in a new tab

(A) RNM counts collected in each study county and (B) radial plots showing the upload and download duration for each carrier for each hour of the day.

As the data collection team was based in Kisumu County, this county had the highest density of data collected (Figure 5A), and thus the most complete temporal plot (Figure 5B). For both Carrier A and Carrier B, there was an observable increase in upload and download duration during the evening hours (7 pm-midnight), with a smaller peak observed during waking hours of approximately 6–7 am. This trend is consistent with previous observations in different areas. In Kisumu County, both Carrier A and Carrier B were observed to have a distinct increase in upload duration compared to download duration, validating the trend observed in the aggregate data.

Spatial analyses

Using the Pharos application, a total of 1,476 unique landmarks were mapped across the four counties (Figure 6A). Individual maps for each landmark type are shown in Supplemental Figure 3.

Figure 6.

Figure 6.

Open in a new tab

(A) Aggregate map of all Pharos-collected landmarks, excluding cell towers. (B) Comparison map showing the OpenStreetMap database schools (black) relative to the Pharos-mapped schools (pink).

Representative photographs of landmarks captured in Pharos are shown in (Figure 7). Of those landmarks, 407 represented primary schools. The Pharos-mapped schools, which were collected by the Mobile Enabled Diagnostics for Schistosomiasis Control Analytics (MEDSCAN) study field team, were compared to a pre-existing dataset of schools obtained from the OpenStreetMap (OSM) project and filtered to only contain primary schools, and the two datasets were found to vary significantly in terms of spatial distribution (Figure 6B) (Humanitarian Data Exchange 2024). Of the combined 822 schools within the 4 study counties, only 46 schools (5.6%) overlapped in space.

Figure 7.

Figure 7.

Open in a new tab

Representative photographs of landmarks of mobile network, schistosomiasis transmission, public health, and cultural significance, taken using Pharos: (A) cell tower, (B) snails, (C) snail-preferred vegetation, (D) water body with fishing boats, (E) a medication dispensary, (F) a super-market, (G) a primary school, (H) a church.

Additionally, the Pharos application landmark data was used to examine the relationship between water bodies, water-dwelling plant life, and intermediate host snail colony presence—all of relevance to schistosomiasis transmission. Using the application, 208 unique water bodies containing 95 instances of water-dwelling plants and 24 snail colonies were mapped. Co-location analysis performed on this data revealed no statistically significant connection between any water-dwelling plant types and likelihood of snail habitation (Figure 8A) but did show statistically significant correlations between locations of water bodies and likelihood of snail habitation (Figure 8B).

Figure 8.

Figure 8.

Open in a new tab

Colocation maps showing (A) the relationship between water-dwelling plants and intermediate host snails and (B) the relationship between water bodies and intermediate host snails.

Seven locations, concentrated around the shore of Lake Victoria were shown to be co-located. Co-location quotients for these locations ranged from 4.48 to 6.94, with p-values less than or equal to 0.04. A co-location quotient of greater than one indicates that categories of interest (water bodies) are likely to have a neighborhood that contains neighboring category (snail presence) features, with higher values indicating stronger spatial association (Leslie and Kronenfeld 2011).

Discussion

In this study, we trialed the mHealth application Pharos across a 4-county area in western Kenya. The Pharos application was designed to assist researchers in preparing for large-scale mobile-based public health projects by field collection of GIS data on landmarks and mobile network stability in the region to assess feasibility for GIS-enabled mHealth applications in the study region. Several applications (e.g., ESRI ArcGIS Field Maps, Kobo Toolbox, ArcGIS Survey123, FAIMS Electronic Field Notebooks), provide robust tools for data collection, geolocation, and mapping, making them widely used in GIS and fieldwork. However, while these applications offer strong support for manual data entry and mapping, they do not automatically capture network metrics, such as timestamped and GPS-tagged signal strength. These metrics are important for assessing mHealth readiness in remote and resource-limited areas. Pharos distinguishes itself by automating the collection of these network metrics without requiring additional user input, in addition to the landmark mapping functionality. The result is a unified tool that supports both epidemiological studies and mHealth infrastructure assessment, particularly for infectious disease surveillance. This unique combination allows Pharos to support spatial epidemiology more seamlessly, linking network readiness directly with location-specific health data.

Our previous work (Scherr et al. 2020) explored network readiness across a significantly smaller area in Macha, Zambia, and several similar trends were observed in this study despite the increase in size of the study area. Although network performance is not the sole indicator of mHealth readiness in an area, it can serve as a good proxy indicator of potential feasibility for ongoing and future studies. Additionally, it can identify areas of need that, when corrected, could improve the likelihood of success of an mHealth intervention. Much of the results focused on two simple, but highly tangible, metrics for this study: file upload and download speeds. The success and failure of data transmission on mobile networks, the speed at which it occurs, as well as the locations where these experiential variables drastically change, is readily perceived by everyday users (Borella et al. 1997; Jacko et al. 2000; Sekikawa et al. 2001; Sears 2003).

Like previous work (Scherr et al. 2020), we were able to identify a primary carrier which would ensure optimal connectivity for an mHealth intervention. Carrier A outperformed the other two carriers, with only a 19.2% failure rate for upload events and 21.2% failure rate for download events. However, one notable factor is the overlapping coverage for Carrier A and Carrier B. In many instances, individuals living in this area own dual-SIM mobile devices, which carry a SIM card for both Carrier A and Carrier B, maximizing the potential of both networks. As shown in Figure 5, despite the lack of a correlation between carrier efficacy and cell tower location, Carrier A and B were observed to have inverse coverage in many areas (Figure S4). This trend emphasizes the potential promise of dual-SIM technology for mHealth success in the region.

Further, certain temporal trends in the data could be validated in comparison to previous work. Similarities in these trends were particularly notable considering the marked differences in study area (e.g., urban center versus rural village, country). The peaks in upload/download duration during the day are consistent with waking and working hours: the small peak in measurement duration during the 6–7 am period indicates increased mobile traffic at a time which correlates well to approximate waking hours for many people, while the peak after 7 pm likely correlates to the period of high internet traffic after individuals leave work and resume steady interactions with their mobile devices.

In addition to the network efficacy study, we utilized the Pharos application to conduct ground-truth mapping across the four-county study region. This study serves as a preliminary mapping effort to support a large-scale schistosomiasis surveillance initiative, with a focus on improving school-based diagnostic testing programs. By identifying community landmarks, such as schools, market centers, and healthcare facilities, Pharos allows researchers to target and engage local populations more effectively. The mapping of local landmarks, especially schools (69.6% of the 1,476 landmarks collected were community hubs, including schools, market centers, churches, mosques, and healthcare facilities), provides critical spatial context to refine point-of-care diagnostic programs. Images of these landmarks will be incorporated into the mHealth application used in our large-scale schistosomiasis surveillance work to help users more easily locate their home on a map during testingdiagnostic testing. Mapping cell towers and analyzing network capacity ensure that mHealth services, which may rely on messaging or data connectivity, are sustainable in these areas. Additionally, the mapping of snails, water bodies, and water-based vegetation has taken on recent importance given the WHO’s recommendation to incorporate snail control as an intervention to disrupt schistosomiasis transmission (Lardans and Dissous 1998; Lo et al. 2018; Sokolow et al. 2018; World Health Organization 2022). Ultimately, mapping of these landmarks in tandem with network performance metrics across our study area will provide helpful context during our schistosomiasis surveillance work, particularly given the focal nature of the disease, allowing us to understand environmental factors that are associated with disease transmission and community infrastructure that may support targeted interventions.

During our mapping of primary schools, a marked difference in spatial distribution compared to the freely available OSM dataset was observed. The OSM dataset tended to have densely populated school information in towns and cities but sparse coverage in more rural areas, whereas the Pharos-mapped schools more adequately covered these rural areas by relying on targeted local knowledge. This may be due, in part, to the fact that the Pharos-mapped schools were pre-selected based on local knowledge of schistosomiasis transmission, targeting schools for the later study that are known to have some level of disease prevalence, and which often tend to be rural areas. Additionally, the OSM dataset is not limited to primary schools, and instead contains schools of all levels and types (e.g. secondary, private) while the Pharos dataset focuses solely on determining the locations of public primary schools within the catchment area. While open-source datasets are a vital part of GIS research, and often provide a starting point for geospatial inquiry with low-cost and easy availability through programs like the Humanitarian Data Exchange, there are also distinct limitations to the information available, especially in more rural areas, as evidenced in this study. In some cases, open-source data is incomplete, inaccurate, or out of date, and it may be difficult to determine how certain datasets were collected (Borkowska et al. 2023). The addition and comparison of open-source data to ground-truth verification using an application like Pharos could be a significant boon to future GIS endeavors and to enrich currently available forms of open-data.

The Pharos application was also used to collect environmental covariate data in relation to one another with respect to possible modes of schistosomiasis transmission. Schistosomiasis is a well-documented environmentally focal disease, transmitted by intermediate host snails of several species (Liang et al. 2018; Wood et al. 2019; Walker et al. 2020; Aula et al. 2021). During this work, field team members were asked to capture images of any snail colonies, inland water bodies, and water-dwelling plant life observed during the primary objective of mapping schools, cell towers, and local landmarks. One difficulty that arose during this mapping process was the photographic capture of snails. Snail colonies are notoriously ephemeral, and the process of predicting their location has been a significant difficulty for many snail-focused studies (Simoonga et al. 2009; Wood et al. 2019; Tabo et al. 2024). Further, certain species of schistosomiasis intermediate host snails such as Biomphalaria choanomphala are deepwater dwelling and require boat-based dredging to be located. Additionally, much of the inland water observed was seasonal, not only making it an unsuitable habitat for aquatic snails, but for plant life as well. Thus, this secondary objective produced a small sample size compared to the primary objective of mapping physical landmarks. However, the data collected still suggests that the Pharos software could be a useful tool not only for real-time mapping of these colonies, but also understanding the environmental factors that influence them in a larger-scale environmentally focused GIS study.

Conclusion

Overall, this research details the deployment and evaluation of Pharos, a mobile GIS- enabled application designed to support large-scale mHealth studies by addressing two critical preparatory needs: mobile network assessment and ground-truth mapping. In the mHealth field, the phenomenon of “pilotitis”—the fatigue and skepticism surrounding mHealth studies due to the limited success of pilots in scaling up—reflects a gap in the foundational work required to ensure that small-scale trials can transition effectively to larger implementations. Pharos helps address this gap by equipping researchers with detailed, contextual data on network readiness and geospatial characteristics specific to each study area. By conducting a preliminary, targeted analysis of network performance and environmental features, Pharos supports scalability and sustainability from the outset, making it more likely that mHealth projects can progress beyond the pilot stage. The application demonstrated effectiveness in mapping a large area of Western Kenya, and its modular design, along with customizable settings, allows for adaptation across various geographic locations and project-specific needs.

The breadth of trends examined within this research stemming from data collection using a singular tool, such as the correlations between plant life or inland water bodies and disease-host snails or the strength and performance of mobile broadband in this area, open the door for further research in environmental epidemiology, mobile health and information technology, and GIS, while demonstrating the importance of leveraging GIS alongside mHealth approaches. Additionally, the creation of a specific database of primary schools frequented by the KEMRI field team in schistosomiasis transmission surveillance efforts, as well as surrounding landmarks of interest for the local communities, may be of significant use to other researchers in this space.

One limitation for this study is the reliance on manual mapping as opposed to less labor-intensive methods such as remote sensing using satellites or unmanned aerial vehicles (UAVs). By utilizing field mapping teams local to the region, the study was able to benefit from the local knowledge and community buy-in implicit in working with known researchers. However, there are many situations in which manual mapping is difficult (e.g. travelling to impassible regions during rainy seasons). This limitation could be easily mitigated by adapting the app for use with more tech-based remote sensing options such as UAVs.

Overall, the availability of an easy-to-use, inexpensive tool capable of examining a breadth of critical variables, which is also easily customizable and deployable, opens many pathways to engage in participatory mapping approaches in future studies. While our current study relied on expert teams to gather data, participatory mapping—which invites local community members to collect data in real time with photographic and spatiotemporal resolution—holds promise for enhancing GIS and addressing data gaps in resource-limited settings (Goodchild 2007; Elwood 2011, 2008; Jones et al. 2012). Participatory mapping, as envisioned in future expansions of the Pharos project, could empower local populations to collect contextually rich data that aligns with the unique geographic and epidemiological characteristic of diseases like schistosomiasis. This approach could mitigate issues of data uncertainty and ephemerality (McLean 2017; Chun et al. 2019; Goodchild 2020).

In future work, we aim to make the Pharos application available to researchers and, eventually, selected local community participants on both local and country-wide levels. This expansion will provide opportunities for mapping landmarks of interest in a variety of spaces, integrating insights from usability engineering and learnability to ensure accessibility to non-specialist participatory mappers (Jones and Weber 2012), extending the scalability of mapping cellular network performance metrics and infectious disease landmarks of interest.

Supplementary Material

Supp 1

Supplemental data for this article can be accessed online at https://doi.org/10.1080/13658816.2026.2641743.

Funding

National Institutes of Health, R01 AI163472.

Biographies

Carson P. Moore, Ph.D., is a Postdoctoral Scholar at Vanderbilt University in the Department of Chemistry. Her research interests include global health, neglected tropical disease diagnostics, and mobile health. Dr. Moore was responsible for conceptualization, data curation, formal analysis, investigation, writing (original draft preparation), writing (reviewing and editing).

Natalie N. Robbins is a Program Manager at the Vanderbilt Institute for Spatial Research. Her research interests include novel applications of GIS, advanced remote sensing, spatial statistics, and geophysics for anthropology and historical preservation. Ms. Robbins was responsible for formal analysis, visualization, writing (reviewing and editing).

Gladys Odhiambo is a Research Scientist and Study Coordinator at the Kenya Medical Research Institute, Centre for Global Health Research. Her research interests include: social science and public health research, qualitative and mixed-methods approaches, neglected tropical diseases and global health. Ms. Odhiambo was responsible for investigation, writing (reviewing and editing).

Kennedy Andiego is a Data Officer at the Kenya Medical Research Institute, Centre for Global Health Research. His research interests are strategic planning for public health interventions, community engagement and management, field data collection and data management. Mr. Andiego was responsible for investigation, writing (reviewing and editing).

Meredith Odhiambo is a Field Technician at the Kenya Medical Research Institute. Her research interests are strategic planning for public health interventions, and community management and engagement. Ms. Odhiambo was responsible for investigation, writing (reviewing and editing).

Fredrick Rawago is a Research Scientist and Study Coordinator at the Kenya Medical Research Institute. His research interests include neglected tropical diseases and global health, project management and logistics. Mr. Rawago was responsible for investigation, writing (reviewing and editing).

Rosemary Musuva is a Research Scientist at the Kenya Medical Research Institute, Centre for Global Health Research. Her research interests include social science research, neglected tropical diseases and global health. Ms. Musuva was responsible for investigation, writing (reviewing and editing).

Maurice Odiere, Ph.D., is a Research Scientist and Head of the Neglected Tropical Diseases Unit at the Kenya Medical Research Institute, Centre for Global Health Research. His research interests include understanding the burden, morbidity and control of neglected tropical diseases, diagnostics and global health. Dr. Odiere was responsible for conceptualization, funding acquisition, project administration, writing (reviewing and editing).

David Wright, PhD., is a Professor at Vanderbilt University in the Department of Chemistry. His research interests include global health, diagnostics, and interdisciplinary science. Dr. Wright was responsible for conceptualization, funding acquisition, project administration, writing (reviewing and editing).

Thomas Scherr, Ph.D., is a Research Associate Professor at Vanderbilt University in the Department of Chemistry. His research interests include mobile health, global health, and data science. Dr. Scherr was responsible for conceptualization, data curation, formal analysis, funding acquisition, project administration, software, visualization, writing (original draft preparation), writing (reviewing and editing).

Data and codes availability statement

Data, codes, and instructions needed to reproduce the reported findings are available at: https://doi.org/10.6084/m9.figshare.29817092. The Pharos application is freely available as open-source software under the MIT License at: https://github.com/tscherr/pharos-open. Researchers are encouraged to use, modify, and distribute the software with attribution to the original authors. In keeping with FAIR data principles (Wilkinson et al. 2016; Hughes et al. 2023), we recommend that non-sensitive data collected using Pharos be made openly accessible through open data repositories.

References

  1. Alsos OA, 2008. Attention and usability issues in mobile health information systems at point-of-care. Studies in Health Technology and Informatics, 136, 877–878. [PubMed] [Google Scholar]
  2. Aula OP, et al. , 2021. Schistosomiasis with a Focus on Africa. Tropical Medicine and Infectious Disease, 6 (3), 109. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Blanford JI, and Jolly AM, 2021. Public health needs GIScience (like now). AGILE: GIScience Series, 2, 1–11. [Google Scholar]
  4. Boonchieng W, et al. , 2021. mHealth technology translation in a limited resources community—process, challenges, and lessons learned from a limited resources Community of Chiang Mai Province, Thailand. IEEE Journal of Translational Engineering in Health and Medicine, 9, 3700108–3700108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Borella MS, Sears A, and Jacko JA, 1997. The effects of Internet latency on user perception of information content. In: GLOBECOM 97. IEEE Global Telecommunications Conference. Conference Record. Presented at the GLOBECOM 97. IEEE Global Telecommunications Conference. Conference Record, Phoenix, AZ, USA: IEEE, 1932–1936. [Google Scholar]
  6. Borkowska S, Bielecka E, and Pokonieczny K, 2023. OpenStreetMap—building data complete-ness visualization in terms of “Fitness for purpose. Advances in Geodesy and Geoinformation, 72(1), e35. [Google Scholar]
  7. Burger PR, et al. , 2024. Mobile health and chronic care: using GIScience to assess health care accessibility among broadband subscribers in Nebraska’s Micropolitan and Rural Areas. Papers in Applied Geography, 10 (2), 96–103. [Google Scholar]
  8. Centers for Disease Control and Prevention 2019. Schistosomiasis infection. Available from: https://www.cdc.gov/dpdx/schistosomiasis/index.html [accessed 6.24.23].
  9. Chun Y, Kwan M-P, and Griffith DA, 2019. Uncertainty and context in GIScience and geography: challenges in the era of geospatial big data. International Journal of Geographical Information Science, 33 (6), 1131–1134. [Google Scholar]
  10. Colley DG, and Secor WE, 2014. Immunology of human schistosomiasis. Parasite Immunology, 36 (8), 347–357. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Colley DG, et al. , 2014. Human schistosomiasis. Lancet (London, England), 383 (9936), 2253–2264. [DOI] [PMC free article] [PubMed] [Google Scholar]
  12. Cumbane SP, and Gidófalvi G, 2021. Spatial distribution of displaced population estimated using mobile phone data to support disaster response activities. ISPRS International Journal of Geo-Information, 10 (6), 421. [Google Scholar]
  13. Déglise C, Suggs LS, and Odermatt P, 2012. SMS for disease control in developing countries: a systematic review of mobile health applications. Journal of Telemedicine and Telecare, 18 (5), 273–281. [DOI] [PubMed] [Google Scholar]
  14. Elwood S, 2008. Grassroots groups as stakeholders in spatial data infrastructures: challenges and opportunities for local data development and sharing. International Journal of Geographical Information Science, 22 (1), 71–90. [Google Scholar]
  15. Elwood S, 2011. Participatory approaches in GIS and society research: foundations, practices, and future directions, 381–399. [Google Scholar]
  16. Free C, et al. , 2013. The effectiveness of mobile-health technologies to improve health care service delivery processes: a systematic review and meta-analysis. PLoS Medicine, 10 (1), e1001363. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Friani G, et al. , 2023. Biological control of biomphalaria, the intermediate host of Schistosoma spp.: a systematic review. Ciência Rural, 53 (4), e20210714. [Google Scholar]
  18. Global Burden of Disease 2020. GBD cause and risk summaries. Avaialble from https://www.the-lancet.com/gbd/summaries [accessed 6.8.23].
  19. Goodchild MF, 2007. Citizens as sensors: the world of volunteered geography. GeoJournal, 69, 211–221. [Google Scholar]
  20. Goodchild MF, 2020. How well do we really know the world? Uncertainty in GIScience. Journal of Spatial Information Science, 20, 97–102. [Google Scholar]
  21. Greve M, et al. , 2022. Overcoming the barriers of mobile health that hamper sustainability in low-resource environments. Journal of Public Health, 30 (1), 49–62. [Google Scholar]
  22. Gryseels B, et al. , 2006. Human schistosomiasis. Lancet (London, England), 368 (9541), 1106–1118. [DOI] [PubMed] [Google Scholar]
  23. GSMA 2023a. Mobile connectivity index. Available from: https://www.mobileconnectivityindex.com/ [accessed 10.4.23].
  24. GSMA 2023b. Mobile industry impact report summary. Available from: https://www.gsma.com/solutions-and-impact/connectivity-for-good/external-affairs/wp-content/uploads/2023/09/2023-Mobile-Industry-Impact-Report-Summary.pdf. [Google Scholar]
  25. Hughes LD, et al. , 2023. Addressing barriers in FAIR data practices for biomedical data. Scientific Data, 10 (1), 98. [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Humanitarian Data Exchange 2024. Kenya education facilities (OpenStreetMap Export), Available from: https://data.humdata.org/dataset/hotosm_ken_education_facilities [accessed 6.11.24].
  27. International Telecommunication Union 2023. Measuring digital development: facts and figures: focus on least developed countries. [Google Scholar]
  28. Istepanian RSH, and AlAnzi T, 2020. Mobile health (m-health): evidence-based progress or scientific retrogression. In: Sternberg RJ & Feng DD, ed. Biomedical information technology. 2nd ed. London: Academic Press. 717–733. [Google Scholar]
  29. Jacko JA, Sears A, and Borella MS, 2000. The effect of network delay and media on user per-ceptions of web resources. Behaviour & Information Technology, 19 (6), 427–439. [Google Scholar]
  30. Jones CE, and Weber P, 2012. Towards usability engineering for online editors of volunteered geographic information: a perspective on learnability. Transactions in GIS, 16 (4), 523–544. [Google Scholar]
  31. Jones CE, Mount NJ, and Weber P, 2012. The rise of the GIS volunteer. Transactions in GIS, 16 (4), 431–434. [Google Scholar]
  32. Jones P, et al. , 2017. The sustainable development goals and information and communication technologies. Indonesian Journal of Sustainability Accounting and Management, 1 (1), 1–15. [Google Scholar]
  33. Kamel Boulos MN, and Koh K, 2021. Smart city lifestyle sensing, big data, geo-analytics and intelligence for smarter public health decision-making in overweight, obesity and type 2 diabetes prevention: the research we should be doing. International Journal of Health Geographics, 20 (1), 12. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Kenya National Bureau of Statistics 2019. 2019 Kenya population and housing census. Nairobi: Kenya National Bureau of Statistics. [Google Scholar]
  35. Konstan JA, Conejo R, Marzo JL, Oliver N 2011. User modeling, adaption and personalization. In: 19th International Conference, UMAP 2011, July 11-15, 2011. Girona, Spain: Springer Berlin Heidelberg. [Google Scholar]
  36. Lardans V, and Dissous C, 1998. Snail control strategies for reduction of schistosomiasis transmission. Parasitology Today (Personal ed.), 14 (10), 413–417. [DOI] [PubMed] [Google Scholar]
  37. Lemaire J, 2011. Scaling up mobile health: elements necessary for the successful scale up of mHealth in developing countries. Geneva: Advanced Development for Africa. [Google Scholar]
  38. Leslie TF, and Kronenfeld BJ, 2011. The colocation quotient: a new measure of spatial association between categorical subsets of points. Geographical Analysis, 43 (3), 306–326. [Google Scholar]
  39. Lewis T, et al. , 2012. E-health in low- and middle-income countries: findings from the Center for Health Market Innovations. Bulletin of the World Health Organization, 90 (5), 332–340. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Liang S, Abe EM, and Zhou X-N, 2018. Integrating ecological approaches to interrupt schistosomiasis transmission: opportunities and challenges. Infectious Diseases of Poverty, 7 (1), 124. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Lo NC, et al. , 2018. Impact and cost-effectiveness of snail control to achieve disease control targets for schistosomiasis. Proceedings of the National Academy of Sciences, 115 (4), E584–E591. [Google Scholar]
  42. Mandahl-Barth G, 1957. Intermediate hosts of Schistosoma. Bull World Health Organ, 17, 1–65. [PMC free article] [PubMed] [Google Scholar]
  43. Mari L, et al. , 2017. Heterogeneity in schistosomiasis transmission dynamics. Journal of Theoretical Biology, 432, 87–99. [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. McCool J, et al. , 2022. Mobile health (mHealth) in low- and middle-income countries. Annual Review of Public Health, 43 (1), 525–539. [Google Scholar]
  45. McLean K, 2017. Mapping the invisible and the ephemeral. In: The Routledge Handbook of Mapping and Cartography. London: Routledge. [Google Scholar]
  46. McManus DP, et al. , 2018. Schistosomiasis. Nature Reviews. Disease Primers, 4 (1), 13. [Google Scholar]
  47. Mechael P, et al. , 2010. Barriers and gaps affecting mhealth in low and middle income countries: policy white paper. [Google Scholar]
  48. Nelwan ML, 2019. Schistosomiasis: life cycle, diagnosis, and control. Current Therapeutic Research, Clinical and Experimental, 91, 5–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Nowak MM, et al. , 2020. Mobile GIS applications for environmental field surveys: a state of the art. Global Ecology and Conservation, 23, e01089. [Google Scholar]
  50. Nyapwere N, Dube YP, and Makanga PT, 2021. Guidelines for developing geographically sensitive mobile health applications. Health and Technology, 11 (2), 379–387. [Google Scholar]
  51. Oliver N, Matic A, and Frias-Martinez E, 2015. Mobile network data for public health: opportunities and challenges. Frontiers in Public Health, 3, 189. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Orr JA, and King RJ, 2015. Mobile phone SMS messages can enhance healthy behaviour: a meta-analysis of randomised controlled trials. Health Psychology Review, 9 (4), 397–416. [DOI] [PubMed] [Google Scholar]
  53. Osei E, and Mashamba-Thompson TP, 2021. Mobile health applications for disease screening and treatment support in low-and middle-income countries: A narrative review. Heliyon, 7 (3), e06639. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Pascoe J, Ryan N, and Morse D, 2000. Using while moving: HCI issues in fieldwork environments. ACM Transactions on Computer-Human Interaction, 7 (3), 417–437. [Google Scholar]
  55. Pearce EJ, and MacDonald AS, 2002. The immunobiology of schistosomiasis. Nature Reviews. Immunology, 2 (7), 499–511. [Google Scholar]
  56. Pires IM, et al. , 2020. A research on the classification and applicability of the mobile health applications. Journal of Personalized Medicine, 10 (1), 11. [DOI] [PMC free article] [PubMed] [Google Scholar]
  57. Rogers DJ, and Randolph SE, 2003. Studying the global distribution of infectious diseases using GIS and RS. Nature Reviews. Microbiology, 1 (3), 231–237. [DOI] [PMC free article] [PubMed] [Google Scholar]
  58. Ryan N, Pascoe J, and Morse D, 1999. FieldNote: extending a GIS into the field. [Google Scholar]
  59. Sama PR, et al. , 2014. An evaluation of mobile health application tools. JMIR mHealth and uHealth, 2 (2), e3088. [Google Scholar]
  60. Scherr TF, et al. , 2020. Evaluating network readiness for mhealth interventions using the beacon mobile phone app: application development and validation study. JMIR mHealth and uHealth, 8 (7), e18413. [DOI] [PMC free article] [PubMed] [Google Scholar]
  61. Sears A, 2003. Simulating network delays: applications, algorithms, and tools. International Journal of Human-Computer Interaction, 16 (2), 301–323. [Google Scholar]
  62. Sekikawa A, et al. , 2001. Does the perception of downloading speed influence the evaluation of web-based lectures? Public Health, 115 (2), 152–156. [DOI] [PubMed] [Google Scholar]
  63. Siedner MJ, et al. , 2012. Optimizing network connectivity for mobile health technologies in sub-Saharan Africa. PloS One, 7 (9), e45643. [DOI] [PMC free article] [PubMed] [Google Scholar]
  64. Simoonga C, et al. , 2009. Remote sensing, geographical information system and spatial analysis for schistosomiasis epidemiology and ecology in Africa. Parasitology, 136 (13), 1683–1693. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Sokolow SH, et al. , 2018. To reduce the global burden of human schistosomiasis, use ‘old fashioned’ snail control. Trends in Parasitology, 34 (1), 23–40. [DOI] [PMC free article] [PubMed] [Google Scholar]
  66. StatCounter Global Stats 2024. Mobile operating system market share worldwide. Available from: https://gs.statcounter.com/os-market-share/mobile/worldwide [accessed 6.10.24].
  67. Statista 2022. Mobile internet traffic share in selected regions 2022. Available from: https://www.statista.com/statistics/430830/share-of-mobile-internet-traffic-countries/ [accessed 10.4.23].
  68. Tabo Z, et al. , 2024. A machine learning approach for modeling the occurrence of the major intermediate hosts for schistosomiasis in East Africa. Scientific Reports, 14 (1), 4274. [DOI] [PMC free article] [PubMed] [Google Scholar]
  69. Thales 2023. How sustainable is the mobile sector? The results are in… Available from https://www.thalesgroup.com/en/worldwide-digital-identity-and-security/mobile/magazine/how-sus-tainable-mobile-sector-results-are [accessed 12.19.23].
  70. Vos T, et al. , 2020. Global burden of 369 diseases and injuries in 204 countries and territories, 1990–2019: a systematic analysis for the Global Burden of Disease Study 2019. The Lancet, 396 (10258), 1204–1222. [Google Scholar]
  71. Walker JW, et al. , 2020. Environmental predictors of schistosomiasis persistent hotspots following mass treatment with Praziquantel. The American Journal of Tropical Medicine and Hygiene, 102 (2), 328–338. [DOI] [PMC free article] [PubMed] [Google Scholar]
  72. Wellcome 2022. Low- and middle-income countries—list. Available from https://wellcome.org/grant-funding/guidance/low-and-middle-income-countries [accessed 1.11.24].
  73. Wilkinson MD, et al. , 2016. The FAIR Guiding Principles for scientific data management and stewardship. Scientific Data, 3 (1), 160018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  74. Wood CL, et al. , 2019. Precision mapping of snail habitat provides a powerful indicator of human schistosomiasis transmission. Proceedings of the National Academy of Sciences of the United States of America, 116 (46), 23182–23191. [DOI] [PMC free article] [PubMed] [Google Scholar]
  75. Woolhouse MEJ, and Chandiwana SK, 1989. Spatial and temporal heterogeneity in the population dynamics of Bulinus globosus and Biomphalaria pfeifferi and in the epidemiology of their infection with schistosomes. Parasitology, 98 (Pt 1) (1), 21–34. [DOI] [PubMed] [Google Scholar]
  76. World Health Organization 2022. WHO guideline on control and elimination of human schistosomiasis. [Google Scholar]
  77. World Health Organization n.d. Schistosomiasis (Bilharzia). Available from: https://www.who.int/health-topics/schistosomiasis [accessed 6.8.23].
  78. Xu W, and Liu Y, 2015. mHealthApps: a repository and database of mobile health apps. JMIR mHealth and uHealth, 3 (1), e4026. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp 1
NIHMS2161662-supplement-Supp_1.docx (3.2MB, docx)

Data Availability Statement

Data, codes, and instructions needed to reproduce the reported findings are available at: https://doi.org/10.6084/m9.figshare.29817092. The Pharos application is freely available as open-source software under the MIT License at: https://github.com/tscherr/pharos-open. Researchers are encouraged to use, modify, and distribute the software with attribution to the original authors. In keeping with FAIR data principles (Wilkinson et al. 2016; Hughes et al. 2023), we recommend that non-sensitive data collected using Pharos be made openly accessible through open data repositories.

RESOURCES