Harmony

The open source Harmony platform is in use 10+ countries. Governments use it to improve public health by making more data-driven solutions. In 2019, the Government of Mozambique used it to coordinate its emergency response to Cyclone Idai, using data to cut the number of new cholera cases from 400 to 0 per day in the most affected province in just 3.5 weeks. Governments on three continents are using Harmony to coordinate their responses to Covid-19.

Zenysis wants to expand Harmony to new countries and make it easier for end users to integrate new sources of data on their own, including by building a front-end tool so that that users can upload Excel and other sources into the platform.

This will help during humanitarian situations and infectious disease outbreaks, when there is a proliferation of data generated, often in silo, and an urgent need to integrate, share, and analyze data.

One of the most significant barriers to realizing the goal of data-driven decision making is fragmentation across data systems. Only when decision makers see data in the context of other data can they identify trends that would otherwise be impossible to see via a single siloed data source.

In the health sector, data fragmentation means that decision makers cannot easily compare supply chain, logistics, and health data to prevent supply chain stockouts and accurately forecast future needs. Human resources and financial data cannot be compared easily against health data, meaning that resources are not allocated as efficiently as possible. Patient data are scattered across multiple systems, so that when a health professional needs to access critical patient information from another health facility, it is difficult or impossible to do. 

Integrating fragmented information systems remains out of reach for is not a well solved problem. Integrating disparate data systems is exceedingly complex due to a lack of shared structure across relevant datasets. Systems can differ in terms of  file formats, infrastructure, levels of aggregation, non-standard units of measurement, among many others. 

This problem exists across industries and in low-income and advanced-economies alike, but the problem is especially acute in low-resourced environments.

The open source Harmony platform seamlessly integrates any type of structured data into a unified view for analysis, harmonizing differences across systems via a back-end integration layer. The result is that end-users can effortlessly access, combine and analyze data from any integrated system for triangulated analysis. The platform then provides decision makers advanced query and visualization tools to explore this data for insights, such as: bar charts, longitudinal time series, structured tables, geospatial analysis, scorecards, heat maps, scatterplots, box plots, and more. 

In addition to enabling analysts to perform custom queries across all integrated systems, the Harmony platform also automates analysis for decision makers, creates real-time alerts, and allows for deep dives on data quality issues. Analysts can save visualizations to shared dashboards so that the most important analyses are available to decision makers on-demand. They can also set alerts to notify decision makers about impending supply chain stock outs or upticks in certain health cases that might signal disease outbreak. Or, they can use the platform’s data quality lab to identify data quality issues across systems. 

The platform’s tools are user-friendly so that end users do not need to have sophisticated data science backgrounds to run highly advanced queries.

The end beneficiaries of our solution are the citizens who live in the countries where we work.

We typically work directly with ministries of health and large national institutes of health to improve data-driven decision-making. 

We embed our teams directly within ministries of health typically anywhere between 1-3 years (sometimes longer) to improve data utilization for decision-making and to make sure that we constantly improve our software so it meets our end users' core needs by taking a Human Centered Design approach.

The solution we provide is a platform governments and large institutions can then use to combine data from any number of formerly fragmented data systems for advanced analysis, often for the first time. Officials have used the platform to design major policies including to: determine where to allocate life-saving health commodities, develop a $120 million nutrition budget, coordinate an emergency response to two back-to-back cyclones, and to keep HIV-positive individuals on the continuum-of-care so that they get the treatments they need (and to receive alerts if someone drops out of a treatment program).

Integration of health and non-health systems will be critical to solving the present Covid-19 challenge as well as proactively identifying and fighting future infectious disease outbreaks. 

The kinds of data sources that decision makers need to proactively identify emerging threats include data about: infectious disease surveillance, climate change, zoonotic information, laboratory data, among many other sources. 

To respond to infectious disease outbreaks, decision makers need access to other types of data, including: hospital, health workforce, supply chain, contact tracing, and anonymized geolocation data.

Integrating and analyzing data is Zenysis' speciality. We support 10+ countries to do this already.

Unlike analytics solutions like Tableau or PowerBi, which give users the ability to visualize a single, siloed data system at a time, the Harmony platform can be used to visualize any number of data from any number of integrated systems at the same time, and combine those data for analysis.

When decision makers can see data in the context of other data, they can surface insights they would otherwise not be able to see when only looking at a single, siloed source.

Additionally, in the health space, there are some open source solutions like DHIS2 which have been used to great effect to integrate some data sources, but DHIS2 is a reporting tool which is not well-suited for integration with other types of information systems that are highly relevant for infectious disease response such as climate change data, supply chain data, and workforce data. 

The Zenysis Harmony platform can integrate any of these structured data systems, including DHIS2 and any number of other systems and data sources.

The Zenysis Harmony solution is a software solution. It is deployable on the cloud or on ministry of health servers.

It is primarily written in Python and it outputs data to an open source Druid database. 

The technology includes a front-end analytics platform as well as a back-end solution that integrates fragmented data systems into a unified database through the automation of data cleaning, data harmonization, and data matching steps.

The solution is in use in 10+ countries, including in: Mozambique, Pakistan, India, South Africa, Bangladesh, Liberia, Brazil, Rwanda, Zambia, Ethiopia, and soon, Vietnam.

Zenysis typically works with national governments inside ministries of health. Our main users are division heads, directors general, and analysts inside governments. We also work with major government implementing partners (which implement on behalf of governments), such as NACOSA in South Africa and HIV/AIDS Alliance in India.

Here is a video demo of the platform: https://www.youtube.com/watch?v=6OeHL9f1IgE&t=6s

Here is a recent Financial Times article describing our work in Mozambique as an example:  https://www.ft.com/content/330b1734-684a-11ea-a6ac-9122541af204

When working with our end users we always start with the impact we want to have. From that, we think about the outcomes we want to achieve, the the outputs we need for those outcomes, and the inputs we need for the outcomes. Answering these questions in this order (starting with the impact question) helps us determine our engineering and user engagement strategies.

For example, in one country, our impact goal was to reduce the percent of people in vulnerable populations living with HIV.

Two example outcomes we needed to achieve in order to have this impact included:

• increasing the percent of ‘at-risk clients’ reached with priority interventions (referred to as the minimum package in epidemiological terms), and

• increasing the percent of overall retention across all HIV-prevention and treatment programs

Two example outputs we needed to achieve in order to achieve these outcomes included:

• increasing the productivity and performance of staff at the ministry of health, and

• increasing their staff utilization of data for decision-making.

Three example inputs we needed to achieve these outputs included:

• improving platform usefulness by doing things like creating dashboards that were highly aligned with staff priorities, and

• providing training around dashboard building and building new features to meet certain core user needs, and

Based on these conversations (and a regular feedback loop), we determined what new features our engineers needed to build, test, and deploy for use among our core users, as well as what data systems we needed to integrate into the platform to facilitate advanced decision-making.

We don't integrate systems for the sake of integrating systems. We integrate systems and enable advanced analytics based on the long-term theory of change we work with our clients to develop together.

Our software serves national institutions that are responsible for providing healthcare to more than 1.3 billion people around the world. 

Because we work with national institutions, the software has enormous potential reach. For example, governments use it to target where to distribute vaccine delivery in low-income countries with populations of tens or hundreds of millions of people. 

Within one year, we expect to work with governments that serve an additional 200 million people.

Within five years, we will work with institutions that serve more than 2 billion people.

Within the next year we will work in 15 countries. We also expect to broaden our reach beyond just health, to include food security, climate change resilience, and post-Covid-19 economic recovery. Within health, we will begin to play a meaningful role towards helping countries achieve their universal health coverage goals, including by linking ministry of finance, ministry of social security, and ministry of health systems.

Within five years, we will be in more than 30 countries, including a mix of advanced economies as well as low- and middle-income economies, addressing a range of social needs, including in health, agriculture, and labor and employment needs.

Most of our projects are donor-funded. Often donors, including bilateral governments and large international organizations, do not move at the speed that our clients would like. Our ability to scale is dependent on our ability to resource our teams, hire engineers, and embed our staff within ministries of health to effect our theory of change. 

Each country presents new challenges as well, especially related to cultures around data use, organizational management, and data management. 

We are actively trying to diversify our funding base, including a mix of private sector capital, foundations, and expanding the range of partners we work through. We are also exploring working with commercial clients to offset some of the costs associated with working in the public sector.

To address country-specific challenges, we typically hire teams from the countries where we work. Our project managers are experienced professionals who have networks and familiarity with the health sectors where they work.