Webinar Recording and Q&A: Accelerating the COVID-19 Response

A live webinar hosted by Zenysis Technologies on April 2, 2020.

This one-hour webinar featured a Zenysis introduction and brief demonstration, an overview of Zenysis’ emergency response experience in Mozambique during Cyclones Idai and Kenneth in early 2019 and a discussion on the phases of COVID-19 response and the importance of data-driven decision-making during the pandemic.

Click to view the webinar on the Zenysis YouTube channel

Q&A Summary

Question: Thanks for the presentation. The data all looks very clean and the systems are all inter-operable. Can you say more about what needs to be done to get the data ready to be used in the platform
Answer: We begin new deployments by surveying the different health systems and understanding their formats, data structures, and the triangulations that may happen between them.

• In order to begin the data integration process, we ask that countries provide read-only credentials so that our software can access existing data system APIs or databases. In the case of Excel spreadsheets, a sample of the data is necessary to prepare the data ingestion pipeline.
• Once a pipeline has been created, it will routinely pull data from the various systems, clean, and harmonize the data sources. Depending on the complexity of entity resolution between data sources, before launching the platform a country might use our Entity Review & Resolution tool to resolve any remaining inconsistencies or open questions. At this point the data is ready to be used in the platform.
• Once data is used in the platform, data quality issues are surfaced through routine analysis or through our Data Quality Lab, which provides real-time data quality scores for each indicator (informed by the WHO standards for data quality).

Question: Has implementation of the Zenysis platform resulted in the removal/obsolescence of existing tools (e.g. a standalone GIS software) within an MOH? How is that decision approached, especially when formal information system agreements are in place?
Answer: Zenysis was created to integrate data sources in the easiest and most cost effective way for countries. Our aim is not to remove existing tools but instead to complement them by integrating their data into our platform for advanced analysis. This means NO additional training costs for countries and no new data collection tools.

Question: How do you ensure that if patients are included in multiple systems that you are aligning records for the correct person and not inadvertently leaving records for an individual as separate or inadvertently combining two separate persons’ records
Answer: The simple answer here is that we use matching algorithms to match individuals across data systems. As a first prize we would rely on national identification numbers or other reliable unique identifiers. But where such unique identifiers do not exist we implement advanced matching algorithms. These algorithms match records across a multitude of demographic factors. The matching algorithms are really flexible in that each demographic factor can have it’s own matching rules. It also matches data in different ways such as exact character matching, nuanced number matching, phonetic matching etc. The algorithm also learns as it goes along and becomes better and better at matching as more data is fed into the system.

Question: Working with different data sources, indicator definitions may vary across the platforms — how does Zenysis aggregate indicator information that is defined differently across those platforms?
Answer: Regarding your question on indicator definitions. When we integrate data across data sources, it is key to empower analysts with enough information to contextualize their analysis and interpret results. This is a mixture of automated and more manual documentation.

• Automated: we automatically generate information regarding frequency of collection, admin level where data is collected, dimensions of analysis that are enabled, and data quality scores.
• Other: when we integrate data and in some cases generate new indicators, data managers in the organizations we work with work to document this integrated data. This usually comes in the form of indicator definitions. These definitions can then be surfaced directly in the analysis tool so users can better understand the data they are accessing. In some cases, there are “live” indicator dictionaries (i.e. Open Concept Lab) which we can connect to and integrate definitions from.
• In some other cases, data is collected with different “units”. For example, we have seen supply chain data come in describing Rapid Diagnostic Tests for malaria described as “boxes of 20” vs. “individual units”. To compare indicators describing RDTs with different units, we standardize units in our harmonization layer. In order to do this in a way that is aligned with use cases, we work very closely with analysts in the MOH and specific programs we are embedded in.

Question: Has the team identified any trends in what types of information are typically needed to harmonize data sources that governments should make sure to prepare if they would like to use this/a similar platform? E.g. nationally standardized health facility list, village list, etc.
Response: There is a small set of key resources that underpin our integrations. These act as “sources of truth” to harmonize all datasets and enable side by side querying of different sources. Depending on the types of data sources we integrate, it is useful to have:

• Master Facility List: this is an official resource maintained by the MOH describing all facilities in the country, and where they exist in the country’s administrative hierarchy. We use the MFL to map all data sources (that use their own internal facility list) to a shared master facility list. If a canonical list doesn’t exist, we work with the partner to build it!
• Population data: population data (usually from Census or a National Statistics bureau) will underpin the calculation of many key indicators.
• Shapefiles: enabling geographical analysis relies on having available “shapes” of the country admin hierarchy.

Please contact us at covid19response@zenysis.com with questions or comments.