Authored By: Rachel Sibande (Digital Impact Alliance), Erwin Knippenberg (Cooper/Smith)
Often, when people think of using network data for development; there are concerns around access and use of personally identifiable information. However, we have found that we can answer development questions using basic insights from network data without the use of personally identifiable information.
There are several elements of Mobile Network Operator (MNO) data that generate basic to complex insights valuable to answer key development questions in a safe and responsible way. For example: questions on understanding population density and migration patterns of people to know where to deploy schools, water points, health facilities and other service delivery facilities can be answered using aggregated location data from network data without accessing personally identifiable information. Questions around income profiles or poverty index to inform food security outcomes can be answered by analyzing airtime top ups, spending data and transactional data on mobile money platforms. Demographic data on gender, age, sex can generate insights on community profiles, lifestyles and usage behavior.
DIAL and its partners have been working to support governments, development practitioners and technology implementers to enhance their efforts in delivering public services. DIAL was commissioned to create turnkey replicable models that uses MNO data to answer development questions in a safe and responsible way. DIAL engages with a cross range of digital ecosystem actors including donors, governments, NGOs, MNOs, Companies and Regulators.
In Malawi, DIAL has partnered with the Ministry of Health as the end user, Cooper/Smith and Infosys (technical partners) to leverage the responsible use of network data to demonstrate how MNO data can be used to inform the placement of new health facilities and direct future investments priorities in health.
The analysis of MNO data, disease burden data from the Ministry of Health, and existing data on location of current health facilities across the country; confirmed that access to primary healthcare remains a challenge in Malawi, with an estimated 45% of people living more than 5 km from a health facility. Without action, this number is projected to increase to 9.7 million by 2023. To remedy this, the Malawi Ministry of Health is evaluating the investment required to build 900 new health posts by 2023.
To inform this potential investment and resulting impact, the Malawi demonstration project has been working since 2017 to develop usable metrics from network data on population density and migration patterns without using personally identifiable information. Using two years of network data containing 12.9 billion records, the analysis counted the density of unique-users in each administrative area and compared it to historical population density.
Shifts in population density over time, (which cannot be directly observed), could then be inferred using shifts in the observed density of unique-users over time. This inference was then validated using preliminary results from the 2018 census. It was established that there is no significant difference between the population projections made by this model using MNO data and the population figures from the 2018 preliminary census report. This result alone validates the credibility of using MNO data to understand population densities. The analysis offered a detailed projection of population movement and population growth patterns, nationwide.
The team then optimized the data to reflect the possible placement of new health posts, specifically in under-served areas with rapidly growing populations. Additionally, the model allowed for inclusion of location specific disease burden and disability adjusted life years (DALYs) data. Based on the projected placement of the new health posts, the model projected that the number of Malawians living within 5-6 km of a health facility could increase from 55% to 95% of the population by 2023.
The culmination of this data demonstration project took place on April 10th, when DIAL together with the Ministry of Health and Cooper/Smith, convened a results dissemination workshop for the Malawi data demonstration project in Lilongwe, Malawi. The gathering consisted of policy makers, development practitioners, academia, regulators and MNOs among others. The workshop was a culmination of a year and a half-long process, of among other things; ensuring regulatory compliance, building the infrastructure to house the data, and demonstrating its potential to inform real-world policymaking.
Dr. Charles Mwansambo, Chief of Health Services in the Ministry of Health and Population in Malawi, stated in his opening remarks,
“The mobile platform has brought huge possibilities in how the delivery of health services can be improved. The data that is generated through the platforms can go a long way in making it easier to plan and to provide quality health services.”
During the workshop, development partners named a number of potential follow-up use cases in the domain of education, disaster relief, and public health, where similar methodologies could be applied. MNO providers and regulators also expressed their openness to such uses but emphasized the importance of developing detailed use cases when negotiating access to the data. The workshop also provided a venue for focused dialogue on the potential to harness MNO data for the public good, nation-wide.
In terms of sustainability, members of the academia expressed their interest in developing partnership agreements with the Ministry and implementing partners in order to build the analytical capacity available in country. As a next step, the project consortium will evaluate the possibility of leveraging support to create a viable data pipeline with these capabilities, that will plug into existing data systems in Malawi.
With this groundwork in place, the consortium is focused on engaging directly with data-users in the government, academia, and civil society to consider replication and scale of this model and to stimulate the appetite for use of non-traditional data such as MNO data to answer development questions. As an initial step, government and development practitioners shall consider replication of this model to answer questions such as where to place schools, water points, and other social services based on gaps in service delivery, population density, and other social dynamics.
We believe that through dissemination and replication, there can be a holistic showcase of how MNO data can be leveraged for the public good. This story is about a unique public private collaboration that has contributed to helping answer critical questions that can improve health access but which is also relevant to multiple sectors. Let us know if you are interested in replicating this model to enhance the delivery of public services and improved decision making in the public sector.