Photo Caption: Dr. Upenytho George Dugum (second from the left), Commissioner Department of Community Health at MOH Uganda, handing over 300 smartphones to the leadership of the Ntungamo district. Photo Credit: UNICEF.
One billion people lack access to healthcare because they live too far from a health facility, so digitally-enabled Community Health Workers (CHWs) have been a lifeline in communities where health infrastructure is lacking. For instance, it's estimated that CHWs prevent a child from dying worldwide every three seconds and evidence indicates that effectively trained, deployed, and managed CHWs also help reduce maternal mortality and the spread of infectious diseases.
CHWs are often the primary point of contact with a health system. Every CHW appointment is an opportunity to learn more about healthcare needs in that village, community, county, or country. From maternal and child health to infectious diseases, CHWs are capturing millions of data points every day in the mobile technologies and digital platforms they use to capture medical records, healthcare needs, and treatment plans.
This digital-first health information can be used to drive decision-making at a macro-level, so that resources are better allocated, which helps more people get the treatment they deserve sooner. However, many governments doubt the credibility of CHW-captured data saying that it has too many errors, and therefore shouldn't be trusted. This is where DataKind steps in.
Building trust in the data is a key part of a responsible digital health solution. This is DataKind’s contribution to a powerful consortium called iCOHS (Intelligent Community Health Systems) designed to support Uganda’s Ministry of Health in its efforts to use and nationally scale digital health solutions to sustainably address bottlenecks in the delivery of community health services. This UNICEF and The Rockefeller Foundation-supported consortium partners DataKind with highly established frontline healthcare organizations, Living Goods, Medic, and BRAC, to develop a community-informed data quality and integrity monitoring toolkit called the Data Observation Toolkit (DOT).
We’re at the stage now where smartphones are being delivered to 200 and 300 CHWs in the Lamwo and Ntungamo Districts of Uganda, respectively, and an intensive training program is underway. On completing the training, the CHWs will be ready to use the Community Health Toolkit - Medic’s digital case management platform - on which DOT will run.
Digital health solutions are only useful if you have trustworthy data. To realize the potential of AI for precision public health and predictive analytics to anticipate, for example, the next health emergency or to prevent bad outcomes, data integrity is a foundational pre-requisite.
Precision public health uses data from traditional and emerging sources to target interventions for populations by person, place, and time, in part with a focus on reducing health disparities.
So, thanks to new investments from the Johnson & Johnson Foundation and Wellcome Trust, DataKind will expand our learnings and development of our toolkit to identify, monitor, and manage inconsistent or problematic data in community health systems. Looking ahead, DataKind will continue tackling critical health data issues, advancing data quality and data trust, and creating digital health solutions. With the Johnson & Johnson Foundation, DataKind will further develop DOT in collaboration with the two largest comprehensive mHealth case management platforms, Dimagi’s CommCare - that supports over 1 million frontline health workers, and Medic’s Community Health Toolkit. In partnership with Wellcome Trust, DataKind will also develop a roadmap for integration of these tools with commercial, NGO, and individual health systems.
Trusted, high-quality data empowers health system managers, leaders, and policymakers to make data-driven decisions about public health policy, proceed with confident investments in frontline health systems, and identify opportunities to optimize health service delivery and patient outcomes. DataKind is looking forward to making this a reality by responsibly sharing these upcoming solutions as both Digital Public Goods and Global Goods that will serve communities all around the world (learn more about data integrity in frontline healthcare here).
As we build toward Digital Public and Global Goods, we need to ensure data ethics are at the forefront of our work. The conversation around data ethics has long been centered in academia and evaluates moral problems related to data, algorithms, and corresponding practices in order to formulate and support ethically good solutions. This recent article, Our Ethics + Responsible Data Science Practices at DataKind, covers many of the typical bias, provenance, and privacy issues.
But what about when the available data is anemic or isn’t available at all? Entire populations can go under- or unrepresented. These are the very same people who would benefit most from ethical data science and AI interventions.
Advances in machine learning and AI allow us to create realistic - but not real - datasets that reflect real world experiences. Recent DataKind projects have found it can be as good or even better for training an AI model than the real stuff.
In his article, Why Nonprofits Should Care About Synthetic Data, DataKind Core Volunteer, Daniel Nissani, explains how synthetic data can help to “bolster datasets that are either too small or have dramatic dataset imbalances”. This approach allows us to do more with limited real data. Now that’s a promising future!
As always, thank you for your support of this critical work!