Good data quality can save lives

In a pro-bono project facilitated by DataKind, a global non-profit that harnesses the power of data science and AI in the service of humanity, we had the great opportunity to join a team of volunteer data scientists to enable the Lwala Community Alliance to monitor the quality of their data.

DataKind teams talented volunteer experts with changemakers to collaboratively design and build solutions to tough social challenges. Founded by a group of committed Kenyans, Lwala works with communities to improve access to quality health care in Western Kenya, reducing maternal and child mortality in doing so. Moreover, Lwala is invested in bringing these services to national levels by supporting the Kenyan government as well as sharing their experiences and research with other institutions.

When frontline Community Health Workers (CHWs) visit their community members and patients, they collect health records and other patient-related data with the help of a data collection tool which is provided by the digital health platform CommCare. This tool is accessible to the CHWs as a mobile app with minimal setup.
In order to provide patients with the best possible health services, reliable data plays a pivotal role. When data is reliable, CHWs can build trust in the tool and the process more easily. Similarly, supervisors and decision-makers need a sound basis for their everyday decisions.

This leads us to the project’s goal, which was to create a solution that enables data analysts and decision-makers to monitor the quality of the collected data. Building on common tools and technologies such as Python, Amazon RDS and Tableau, we implemented various data checks that run in regular intervals and whose output can be explored in a customised dashboard.
This interactive visualisation empowers users to identify trends and concrete or potential data issues early and take action quickly – thereby improving data quality for everyone and potentially saving lives.

Read more about the project and its background on the DataKind blog and in the case study by the Johnson & Johnson Center for Health Worker Innovation.