Health Campaign Optimisation Tool

The problem: Due to the complexity of optimizing logistics to deliver health campaigns or to respond to pandemics in low-resource settings (caused by vehicle capacity restrictions, supplies carrying capacity, fuel availability and poor connectivity among other challenges) Riders have identified an acute risk of the current approaches resulting in the hardest-to-reach communities being underserved.

The solution: Implement the “Health Campaign Optimization Tool”, an adapted version of a proven route optimization tool designed to generate operational efficiencies in the deployment of Riders’ fleet and safeguard remote communities from being underserved. 

Positive change if scaled globally: As the tool will be built open-source, Riders and DataKind will actively promote scaling beyond Riders’ operation countries to improve the effectiveness of other campaign-based responses for COVID-19, HIV, tuberculosis, malaria and expanded programs on immunizations so that underserved communities in all geographies are safeguarded against exclusion.

Like most countries, Lesotho relies on campaign-based delivery to extend the coverage of its health system when a time-limited, at-scale health intervention is needed - such as their COVID-19 response. However, while it might be one of Africa’s smallest countries, Lesotho is one of the most challenging when it comes to healthcare service delivery. The majority of the nation's 2 million residents (72%) live in remote and sparsely populated areas that are difficult to access given the poor road infrastructure outside the capital and rugged, mountainous terrain. 

Robust and effective health campaigns and pandemic responses require coordinated efforts amongst stakeholders, senior government support, and innovative approaches to long-standing challenges. 

While Riders has developed a unique award-winning system by expertly selecting appropriate vehicles and by training couriers to handle samples in Lesotho, optimizing logistics in such low-resource settings is made very difficult by long distances over difficult terrain, vehicle capacity restrictions, fuel availability, poor connectivity and data systems, and more. Manual scheduling systems that try to take all of these factors and more into account are often inefficient and result in the hardest-to-reach communities being underserved.

DataKind will enhance Riders resource capacity planning processes by applying data science and AI technology to Riders existing workflows, so they can increase service coverage and improve operational efficiency and effectiveness.

Health campaign delivery can be modeled as a capacitated vehicle routing problem: optimising routes for a fixed number of vehicles with a carrying capacity, that are required to visit a set of locations in a prescribed catchment region within a defined time period. 

DataKind has worked with partners in many contexts on components of the resource-allocation problem (how best to distribute vehicles, customers, and routes to reduce costs, time, or environmental impact) and the “maximal-coverage problem” (modeling optimal location or routing patterns) with the aim of helping organizations scale their services at the lowest possible resource expenditure. In 2019, DataKind built and deployed a route optimization tool for an iNGO in Haiti, to help local staff transport and exchange no-flush toilets within select communities. The tool produced many operational efficiencies including reducing the number of required trips by 13%, lessening the partner’s operating costs and environmental footprint (key objective). The tool offers ongoing benefit to the partner organization, as adding additional stops triggers an automatic update to optimized routes.

Riders for Health began nationwide operations in Lesotho in 1991, mobilising outreach health workers on motorcycles and operating a nation-wide sample transport courier system.  Our office and vehicle maintenance workshop are based in the capital Maseru, providing technical services for vehicles running across Lesotho’s 10 districts. Our countrywide sample transport service has been hugely successful in reducing turn-around times for test results and was the model for scaling to Nigeria, Malawi, and Liberia. Riders for Health Lesotho is managed and staffed by Lesotho nationals. They have a very close relationship with the Ministry of Health and work with numerous partners to help them achieve their health outcomes; including CDC, The University Research Company (URC) and Clinton Health Access Initiative (CHAI).

The HCOT will be developed and deployed in Lesotho drawing on our extensive operational and local knowledge. Lesotho has a population of approximately 2 million people, 72% of the population live in rural mountainous challenging terrain, with extremely limited road infrastructure extending to the last-mile communities.

In order to deliver an effective health campaign or respond to a pandemic, all the population deserve the right to be treated equally and be provided access to the initiatives and solutions set up to strengthen the nation's health. The HCOT, through its innovative use of data science, vehicle tracking and AI, will be able to ensure all of the population, no matter where they are, are given the same access to healthcare.

The performance of health campaigns varies dramatically, and many never realize their intended success (e.g. only 47% of measles campaigns and 38% of Vitamin A campaigns reach their intended coverage targets, with the most vulnerable communities most likely to be underserved). A vehicle routing tool to optimize for field coverage and/or for resource allocation could improve campaign coverage. Such tools are not unique, but they are costly and intended for the for profit sector. MapQuest, Google Maps and MapMyRide allow users to customize their travels as they want, from shortest distance to ‘avoid highways’. However, these tools rely on systems that are more commonplace in High-Income Countries including; accurate digital mapping, clearly defined address systems, and access to cloud computing data for route plotting and sharing. 

The HCOT is designed to address the specific requirements of, and challenges associated with, running health campaigns in Low- and Middle-Income Countries (LMICs).  

The project team will use geospatial data, available through OpenStreetMap, and partner-collected information on facility locations, transport requirements, and transport vehicle size and staff to design and create a solution which offers transformative improvements immediately while also offering long-term sustainability for continuous impact and value. It does not repeat conventional approaches but rather leverages local expertise to inform the construction and adoption of a data science powered optimization technology. The HCOT will fundamentally change health campaign planning by focusing on resource-expense reduction to be able to more efficiently serve the largest number of people. 

This project builds on the application of numerical optimization, specifically around different types of vehicle routing problems (VRPs) using multiple open source software technologies and custom code to provide easy and unique access to these powerful optimization tools. In this project, we focus on “rich” VRPs, this variant is characterized by vehicles potentially having varying capacities and costs, accessing divergent start and end points and multiple depots. 

This project primarily leverages three different open-source software projects. 

The first, OpenStreetMap (OSM), provides the most up-to-date and free information on local roads for locations across the globe and allows for real-time edits and integration of local knowledge (e.g. a route being impassable). 

The second, Open Source Routing Machine (OSRM), is a routing software that computes distances and times between points based on the OSM data, formulating potential routes. 

Third, to account for the complexity of numerical optimization, this project integrates Google OR-Tools, a free, open-source software that facilitates vehicle routing, integer and linear programming, and constraint programming. 

By leveraging existing open-source technology and known access to smartphones, the team has created a workflow to excel in LMICs for frontline resourcing through custom software creation linking the open source technology components and producing offline-friendly maps and instructions to reduce the resource cost for facility visits.  To execute, the user provides information on vehicles, staffing, and locations for the routes and the Riders team uses two technologies to collect and share this information: vehicle tracking systems and mobile data systems

This project utilizes open source and open access technologies. See Bloomberg Data for Good Exchange preprint (forthcoming in 2020) for an academic paper on DataKind’s integration of these technologies in LMIC contexts. Additional documentation by SOIL Haiti and Datafoss.  

(1) OpenStreeMap (OSM) is a free map of the world with data made available under the Open Database License. OSM allows communities to make maps to serve practically any interest group. OSM has over 6million geographically diverse registered users due to emphasis of local knowledge and ground truth in the process of data collection: users collect data using manual survey, GPS devices, aerial photography, and other free sources. 

(2) Open Source Routing Machine (OSRM) is a free network service under the permissive 2-clause BSD license. It is a C++ implementation of a high-performance routing engine which combines sophisticated routing algorithms with road network data from OSM to compute and output a shortest path between any origin and destination within a few milliseconds. The project GitHub has been forked over 2,000 times and has over 100 contributors. 

(3) Google OR-Tools is open source software for combinatorial optimization which seeks to find the best solution to a problem out of a very large set of possible solutions. In most cases, optimization problems have a vast number of possible solutions—too many for a computer to search them all. To overcome this, OR-Tools uses state-of-the-art algorithms to narrow down the search set, in order to find an optimal (or close to optimal) solution.

As part of our digital transformation, Riders assessed various data science focused organisations, DataKind was selected as the most suited and experienced to achieve our aims. 

Since being founded in 2012, DataKind has completed 300+ projects that boost the impact of a social change organization through the thoughtful application of cutting-edge digital technologies. DataKind’s theory of change is not merely to make a frontline partner organization more efficient - though that does happen - but to leverage new computational approaches like machine learning and AI to better address complex social issues. 

 Each project requires the same six components (inputs) before it can launch: a data literate mission-driven social actor addressing a pressing humanitarian challenge, a detailed project plan centered on a pressing organizational need, the datasets required by the proposed solution, technologists to build the deliverables identified in the project plan, subject matter experts to contextualize the issue space so the project is socially responsible, and project funding to underwrite project expenses. With these six inputs in place, DataKind can follow its established ‘DataCorps’ program (activity) to create the deliverables stated in the project plan (outputs) - be they an algorithm or an analysis - so the social actor is more efficient and can better serve their cause for greater impact (outcomes). Beyond helping a frontline organization advance their mission, another outcome of a centering a long-term partnership around the use of technology for social impact is that the partner organization increases their data maturity and adopts a more data-driven approach to their work. 

For this project, the outputs will include the HCOT, which will allow operation managers at Riders to optimize campaign logistics for improved health outcomes. Some campaigns might prioritize population coverage (e.g. a vaccination drive) where others might prioritize efficiency (e.g. COVID-19 education). The HCOT will allow Riders to ensure that ‘last mile’ communities are not underserved by health campaigns because available resources (including time, money, transport capacity) have been extinguished or used inefficiently. To ensure HCOT success and continuous improvement, an impact evaluation will also be designed during the project design stage.

On deployment of the HCOT, Riders will reach approximately 2 million people across Lesotho. Upon scaling to the other countries where Riders operate, In 1 year: ~20 million (scale to Nigeria). In 5 years: ~47 million upon scaling to other Riders operation countries.

Riders and DataKind will promote scaling the HCOT beyond Lesotho to improve the effectiveness of other campaign-based responses for COVID-19, HIV, tuberculosis, malaria and expanded programs on immunizations so underserved communities in all geographies are safeguarded against exclusion. Riders will expand the deployment of the HCOT:

(1) Nationally: Riders will collaborate with the MOH Lesotho to expand use of HCOT to maximise the reach and impact of all health campaign-based service delivery programs in the country - including for HIV, TB, malaria and Expanded Programs on Immunizations campaigns

(2) To Nigeria where Riders and DataKind have been working together since 2019 to drive operational efficiencies in Riders’ sample transport programs using data science and AI. Together, we will expand the deployment of HCOT to 18 states where operations include 131 sample transport couriers that serve 17 laboratories, 1637 PHCs and a total population reach of approximately 20 million people.

(3) To other countries where Riders operate sample transport, including Liberia, and Malawi.

(4) To new countries across Africa where sample transportation is currently underserved or operationally challenging.

Furthermore, the applications of the HCOT stretch far beyond health campaigns for example, home deliveries of medication to vulnerable patients, supply chain management, fleet operations and more. Based on the outcomes of the impact evaluation, DataKind will refine the HCOT to meet the standards and specifications required to designate it as a Global Good, which includes it being open-source and open-access and support other frontline health organizations adapt, test and implement this tool. 

Financial support to build and develop the HCOT.

Both Riders and DataKind are approaching various health and technological focused organisations as well as philanthropic foundations to support the costs required. We have already received 100 vehicle trackers (value of $7,500) towards this project from GeoTab, headquarters in Canada.

Riders have a range of staff working on this solution including sample transport couriers, workshop technicians and office based operations team members in Lesotho. This national team is coordinated through the Research, Development and Digital Lead based in Riders international team. We anticipate as the project progresses there will be additional staff involved in Lesotho as the project moves through its different phases of development.  DataKind is resourcing three technical team staff members part-time to this project, as well as a part of a portfolio manager’s time and a team of 4-6 pro bono data scientists working part time.

Riders has been operating for over 30 years mobilising health workers across Africa, continuously improving the methods of service delivery. We are currently going through a digital transformation of our operations, moving away from traditional paper-based data collection and introducing new and innovative technologies to further enhance our work to strengthen national health systems. Other tech-based solutions we are currently working on are app based data collection for all field based and maintenance workshop staff, vehicle tracking, optical character recognition for track and tracing of patient samples, and barcoding of vehicle replacement parts for stock management. 

Since its launch in 2012, DataKind has delivered over 300 data science solutions to mission-driven organizations working on the Sustainable Development Goals. Across a range of programs, DataKind has established a project process for identifying data science-able problems and co-creating digital solutions to help the social sector partner amplify their impact. DataKind has specifically worked on creating tools for front-line service delivery and route optimization in LMICs since 2018, with public review and demonstrations of the work in 2018. 2019, and 2020. 

Riders partners with multiple partners across its operating countries see here: https://www.riders.org/partner-with-us/ and here: https://www.riders.org/partner-with-us/support-our-work/ Our partners are a blend of local and national health related organisations, ministries of health, international institutions and organisations, philanthropic and social enterprise donors, and motorcycle focused organisations, such as the FIM and their work in MotoGP.

DataKind partners with social change organizations with strong theories of change and clear pathways to positive impact. DataKind is an iNGO with chapters in four countries, and volunteers and partners across 98 countries. Our partners include large foundations eager to offering capacity building opportunities to their grantees, such as the Rockefeller Foundation and the Gates Foundation, health platform actors such as Medic Mobile, care delivery providers such as Muso and Living Goods, government entities such as New York City’s mayor’s office and corporations such as Bloomberg, Microsoft, Google and others. A sampling of DataKind’s project partner organizations can be found here: https://www.datakind.org/blog, and it’s main project funders can be found here: https://www.datakind.org/support-us 

Across the world, social entrepreneurs are taking on some of the world’s biggest and most challenging problems. By recognising and building the talents and abilities of the people living in developing countries, and applying their own innovations and experience, social entrepreneurs are able to create enduring social benefits. Riders is one of these organisations.

Riders’ programmes are designed to become financially self-sustaining in the long-term. Our model is flexible, and means that Riders’ programmes are replicable and can work just as effectively for a small fleet of two or three motorcycles as it does for a government ministry’s fleet of hundreds of two and four wheeled vehicles. Riders work with other social entrepreneurs allowing communities to work together and transfer their skills. We have been recognized for our social enterprise work by the Skoll foundation, Schwab foundation, Ashoka, and Impact 2030. Our services include: sample transportation, outreach healthcare, medical supply chain logistics, fleet management, and rider and driving training.

Operating as a nonprofit ‘tech for social impact’ consultancy, DataKind creates digital solutions which are tailored to boost the specific needs of a mission-driven partner and the communities they serve. Outputs vary from analysis to algorithms, dependent on the partner's requirements.

Riders receive income from a range of sources; contracted delivery of services to health related organisations such as CDC, URC, and the Elizabeth Glaser Paediatric Aids Foundation in Lesotho. We also receive regular funds from our UK based fundraising partner, Two Wheels For Life, for specialist projects including the Digital Transformation project this solution is a part of.

With regards to fleet management for ministries of health, Riders secures loans from socially focused capital investment institutions and banks to pre-fund the procurement of the vehicle assets. The ministry of health repays the loan for the fleet and services delivered over a 5 year time period. This approach to funding Riders has led to an award by the Skoll foundation for social enterprise in 2006.

DataKind’s B2B programmatic offerings are supported by a mix of foundation grants, corporate sponsorships, and individual philanthropy. As long as there is an increasing demand for data science in the social sector, but a limited supply of affordable data science capacity, there will be a need for a ‘match-making’ organizations with access to a community of pro bono data scientists - such as DataKind. DataKind’s operations are supported by a multi-year grant from the Rockefeller Foundation and the Mastercard Center for Inclusive Growth. As a leader in the ‘data and AI for social impact’ space, DataKind leverages its expertise and experience to earn income from others eager to engage in the space - from supporting open call challenges with Google.org and data.org, to Thought Leadership platforms.

Support costs to developing and deploying the HCOT project.

Other health, transport, and data science organisations.

The work carried out by the sample transport couriers and the team behind them is extremely important and should be recognised. They safely handle hundreds of HIV, TB, MDR-TB, COVID-19 samples on a daily bases. They travel long distances each day to ensure no patient or health facility, no matter how remote, receives a professional and friendly service. The communities where we work see them as heroes due to the extreme lengths they go to for delivering samples and returning the results.

Through the use of the innovative technology which will be used to strengthen the health system, initially in Lesotho, but scaling the solution will lead to benefits for anyone who uses the tool globally.