Battuta Radar

Primary healthcare centers (PHCs) are often the first point of contact for patients. There, patients are diagnosed for a disease, triaged to a speciality medical center, and administered medication. Access to such centers is dependent on the availability of public information about their physical location and the services they offer (e.g., blood tests). However, this information is often out-of-date, incomplete, and lacks sufficient detail on core services. This makes it challenging to reliably quantify primary health care performance indicators such as health center density (per 100,000 population).

Most recent estimates of health center location and density by the World Health Organization (WHO) date back almost a decade to 2013 and involve manually coordinating with national ministries of health, a laborious process which introduces a time delay in the reporting of information. Further, there may be difficulties in identifying impromptu or short-term health centers that arise to deal with acute health crises, akin to those during the Covid-19 pandemic. While existing crowdsourcing technology such as that provided by healthsites.io offers to provide more up-to-date information about the location of health centers, it is prone to human errors and inconsistencies in reporting. Further, other applications such as WHO’s Global Health Facilities Database have yet to be operationalized.

As a pilot study, we focus on the country of Saudi Arabia. Whereas official statistics from the Ministry of Health (MOH) date back to 2020, demonstrating a 2-year lag, unofficial statistics retrieved through crowdsourced data are incomplete and inconsistent, capturing only 35% of all health-like utilities officially reported by the MOH. 

Without up-to-date, detailed, and accurate information about health centers, patients are likely to waste limited resources traveling to, for example, non-existent health centers, delaying access to care, and exacerbating patient outcomes. Ultimately, our solution seeks to reliably address a key question: where are the health centers?

We have developed a platform that reliably detects and tracks the evolution of the location of health centers both spatially (e.g., within the borders of a country) and temporally (e.g., over years). Specifically, we use artificial intelligence (AI) technology to detect health centers based on aerial satellite imagery and geolocation data (i.e., latitude and longitude coordinates). Our AI system is trained on existing data to identify both the presence and relative size of health centers. Such an approach lessens the dependence on (a) timely data collection strategies that rely on the involvement of national ministries of health and (b) potentially inaccurate information from crowdsourced platforms. Importantly, when combined with publicly-available population-level statistics, our AI system can improve the reliability of existing primary health care performance indicators such as health center density (per 100,000 population).

Our solution impacts both patients and the broader healthcare community delivering value at multiple levels. It targets patients seeking easy and reliable access to healthcare, otherwise referred to as last mile health. By providing such patients with up-to-date, complete, and accurate information about the physical location and core services offered by nearby health centers, we streamline their access to care, making it more affordable, avoiding unnecessary delays in treatment, and improving their standard of living.

As for the broader healthcare community, our solution impacts (1) policy makers (e.g. at the national, regional and global level) to help inform healthcare financing to improve the financial management of PHC centers, and strengthen healthcare sector governance and accountability through assessing gaps in service delivery across public and private healthcare facilities, (2) prospective PHC providers (e.g. job seekers such as doctors, nurses, within respective communities), and (3) non-governmental organizations, civil society organizations, development partners and others whose objectives are to produce PHC-oriented evidence and innovation using grants and donor funds.

While acknowledging the complexities of access to healthcare such as the availability of affordable transportation, and the ability to take time off of work, as well as other factors such as patient demographics (i.e., age, gender, marital and/ or socioeconomic status) which highlight inequities and may impact the utilization of such centers, we nonetheless believe our solution is a first-step and prerequisite to access.

To demonstrate the feasibility of our platform, we have developed a minimal viable product for health centers located in Saudi Arabia. Although we have initially focused on this country, our platform easily scales across communities from around the globe. This is primarily due to the abundant availability and diversity of aerial imaging.  

Saudi Arabia’s health reforms prioritize tackling the increasing noncommunicable disease burden by prioritizing PHCs, centering it as the core of the newly proposed Model of Care. The country is adopting three pillars that will lay the foundation for successfully achieving this: (1) facilitate access to health services, (2) improve the quality and efficiency of health services, and (3) promote health risk prevention. Saudi Arabia has already started shifting focus and investment from secondary and tertiary healthcare facilities toward reforming and restructuring primary healthcare, aiming to realize these goals. We plan to coordinate closely with the Ministry of Health and, in alignment with the Kingdom’s Vision 2030, we hope to create a paradigm shift in how health performance indicators are quantified and acted upon. 

Originally from the broader Middle East but having grown up in Saudi Arabia in a multi-generational family of physicians, members of the team are acutely aware of the unique challenges faced by its healthcare system. Collectively, we have spent 40+ years in the country and region, allowing us to leverage our localized knowledge to expand this solution beyond the Kingdom. In the near future, we aim to expand our operations to the broader Middle East and North Africa (MENA) region as part of our quest to improve the accessibility of care to all patients.

We hope this Challenge can help us overcome two primary constraints. The first is access to more complete geolocation data on a global scale. Through this Challenge, we aim to partner with organizations and academic researchers who have collected and curated accurate geolocation data (e.g., latitude and longitude) for healthcare institutions. Given the complexity of this data, we appreciate that such partnerships are likely to be country- and region-specific. The second challenge is that of soliciting feedback from organizations and individuals who are likely to be end-users of our platform. By partnering with organizations whose goal is to calculate and update health performance indicators, we can better gauge the utility of our platform and iteratively improve upon its design and implementation. We hope to identify and make connections with such end-users through this Challenge. 

Our solution provides a significantly improved approach to reliably detecting and tracking the evolution of the location of health centers both spatially (e.g., within the borders of a country) and temporally (e.g., over years). Currently, no such comprehensive approach exists that can be scaled up. During our research, we found pockets of work done for certain parts of the world e.g. Sub-Saharan Africa, but once we dug into the data, it was erroneous. 

We use AI technology to detect health centers based on aerial satellite imagery and geolocation data (i.e., latitude and longitude coordinates). Our AI system is trained on existing data to identify both the presence and relative size of health centers. 

Such an approach lessens the dependence on (a) timely data collection strategies that rely on the involvement of national ministries of health and (b) potentially inaccurate information from crowdsourced platforms. Importantly, when combined with publicly-available population-level statistics, our AI system can improve the reliability of existing primary health care performance indicators such as health center density (per 100,000 population). 

In the next year, as part of Phase 1, we anticipate to come closer to achieving our impact through new partnerships stemming from this Challenge, as well as strengthening existing local country partnerships (Saudi Ministry of Health) based on our pilot, that will serve to complement our improved AI solution.

In the next five years, our desired impact is to improve patient access to healthcare in regions where health center information is incomplete and out-of-date. To achieve this impact, we will build upon our Phase 1 findings, scale our platform to additional countries, and continuously monitor our solution’s effect on our outlined indicators of success (e.g., reduced travel times to health centers by patients).

1) Reduction in time wasted by patients searching for health centers

2) Reduction in travel times to health centers

3) Quality of information as perceived by patients in terms of its reliability and utility, by respective Ministries of Health to complement their work, based on a survey scale

4) Number of patients who have accessed PHCs based on the data we provide

5) Numbers of health centers added to official statistics based on our data

Impact (long-term): Improving patient access to healthcare in regions where health center information is incomplete and out-of-date. Key indicators of success will include a reduction in time wasted by patients searching for health centers and a reduction in travel times to health centers. 

As it pertains to access to care, there are some factors beyond our control. These include the ability of patients to afford transportation to and from such centers, whether the services that are on offer are adequate for the patient’s medical condition, amongst other socio-economic factors. From a behavioral standpoint, patients and other end-users alike must gain trust in the information provided to them by BattutaRadar. We envision such trust being gained through educational initiatives informing users of the mechanism used to generate the information being provided to them, and over time, as users begin to act upon this information and truly discover a benefit to them. 

Outcomes (medium-term): Patients and both public and private agencies will have a more reliable estimate of the density of health centers in a particular location. For patients, this will help to streamline their decision-making on access to healthcare. For public and private agencies, this will feed into their governing and funding approaches in the sector. Key indicators of success will include the quality of information as perceived by patients in terms of its reliability and utility, by respective Ministries of Health to complement their work, based on a survey scale, the number of patients who have accessed PHCs based on the data we provide, numbers of health centers added to official statistics based on our data.

Outputs (short-term): By providing patients with more up-to-date and complete information on the whereabouts of nearby health centers, they are better equipped to make decisions related to their access to healthcare. Key indicators of success will include the quality of information as perceived by patients in terms of its reliability and utility based on a survey scale, and the number of patients who have accessed PHCs based on the data we provide. 

Activities: Our solution, BattutaRadar, aims to provide a more reliable estimate of health center density by leveraging artificial intelligence with aerial imaging and geolocation data to detect health centers spatially (e.g., within a country) and temporally (e.g., over time). Key indicators of success will include developing an AI system whose accuracy surpasses a predefined level (e.g., >80%). 

Inputs: As part of our solution, we will invest in resources including aerial imaging data (from satellites), geolocation data (e.g., latitude and longitude) from official and crowdsourced databases, and in the development of AI systems capable of detecting health centers. Key indicators of success will include the total number of health center entries in our geolocation database and the completeness of our aerial imaging database.

Our core technology is centered on AI technology to detect health centers based on aerial satellite imagery and geolocation data (i.e., latitude and longitude coordinates).

Currently, PHC data is collected by ministries of health and/ or is crowdsourced by citizens of respective countries. For the former, the driver is to inform official statistics and governmental policies in the healthcare sector moving forward, including the implication on fiscal budgets. For the latter, the driver likely emanates from interest in contributing to the public good repository of providing such data to aid communities in locating PHCs. 

Having grown up in MENA, with over 40+ years spent living and working in the region, our approach to DEI is driven by our multi-cultural team and diverse skillset that allows for cross collaboration in areas of engineering, (health) policy and economics. Battuta Health's policy is to take a proactive approach to expanding its team in a way that harnesses different and dynamic perspectives. 

We are committed to improving the access to care by patients irrespective of their geography, background, gender, ethnicity and other factors. Our solution, by default, is likely to most benefit those who have been traditionally marginalized in terms of their access to care.

Our value proposition is centered on improving patient access to healthcare in regions where health center information is incomplete and out-of-date. 

Our intervention is in the form of a service in which we provide end-users with complete, up-to-date, and accurate information about the location of health centers. 

Our partners are primarily the ministries of health in respective countries and organizations / academic institutions who have or are in the process of curating healthcare-specific geolocation data. Involving such partners in our solution will facilitate the initial flow of data required for Battuta Radar. 

Our key activities include organizing the various data types that our solution depends on (aerial imaging and geolocation data), developing and maintaining the AI system which tracks health centers, and providing our stakeholders with the requested information. 

Our segments are the public policy makers, patients, and potential job seekers in the healthcare sector. They benefit from our solution primarily through more complete, up-to-date, and accurate information about the location of health centers, thereby better informing their respective decision-making. Our customers are likely to be government entities and organizations who ordinarily collect and update, through traditional means, the health performance indicators of specific geographical regions. 

Our revenue will be in the form of a hybrid model. This includes initial upfront funding (in the form of grants from the government we have established a relationship with) and a more sustainable two-tiered subscription model thereafter. In Tier 1 (basic) pricing, we provide access to the location of health centers in a particular geographical location within a predefined radius. In Tier 2 (advanced) pricing, we provide access to how such health centers have evolved over time, and an exhaustive list of the type of medical services they offer.

Our cost structure constitutes the level of effort required to gather, curate and maintain the repository of data, as well as refine the AI technology on the backend. As we scale, we will likely need more capacity to do these key activities. We will also need to add two team members: the first will be responsible for commercial outreach activities who will be interfacing with the governments to ensure these feedback loops are accounted for, and that our partnerships are working successfully, and the second will interface with our end-user patients as ultimately the success of this channel is important to incent respective governments to partake in our revenue model. This is because the knowledge that more accurate and reliable access for patients to health center locations will benefit the government’s expenditure on the sector and alleviate secondary and tertiary care burdens.

As BattutaRadar establishes its footprint, the ultimate revenue model will be to charge a fee for a proposed multi-tiered service. Ideally, bulk funding will emanate from our government partnerships in the form of grants and/ or paid subscriptions. Subsequently. our tiered offering will be based on the quantity and detail of the information we provide to end-users. For example, in Tier 1 (basic) pricing, we provide access to the location of health centers in a particular geographical location within a pre-defined radius. In Tier 2 (advanced) pricing, we provide access to how such health centers have evolved over time, and an exhaustive list of the type of medical services they offer.

As a start-up, our initial funding base has been personal contributions for working capital. We have also identified 2 grants for which we anticipate to apply to support our efforts in the coming 6 months.