Artificial intelligence for good: using machine learning to close the immunization gap
By Drew Arenth, Chief Business Officer, macro-eyes
Mar 5, 2019
Posted in Uncategorized
Photo: macro-eyes. The macro-eyes team poses with Hassan Mtenga of the BID Initiative, and Dr. Dafrossa Lyimo, Program Manager for Immunization and Vaccine Development in Tanzania.
Electronic immunization registries (EIRs) offer a comprehensive solution to closing the immunization gap. With more than 1,600 health facilities using data quality and use interventions and half a million children registered in EIRs in Tanzania and Zambia, the BID Initiative is on the frontlines of the digital revolution. Thanks to strong government leadership, Tanzania and Zambia maintain some of the most robust vaccine data sets on the African continent. Step aside Silicon Valley, Zambia and Tanzania are ushering global health into the era of artificial intelligence (AI).
In recognition of this government leadership, macro-eyes, a Seattle-based machine learning company with a mandate to increase access to care, was eager to learn from Tanzania’s and Zambia’s vaccine data sets. The Bill & Melinda Gates Foundation awarded macro-eyes funding in 2017 to design and test a predictive supply chain for vaccines. The macro-eyes team worked with the BID Initiative and the Government of Tanzania to access the Tanzania Immunization Registry (TImR) data across more than 700 health facilities and nearly 500,000 patients.
The global need is clear: vaccination rates are slowly rising, but immunization inequity is rife. Some populations have vaccination rates nearing 100 percent; others have rates below 50 percent. Equity is achieved through efficient and precise deployment of resources to both the supply-side and demand-side of vaccine delivery. The secret to vaccinating these hard-to-reach populations is locked inside the data. In this case, that data is in TImR.
Photo: macro-eyes. The individual-level data available through Tanzania’s electronic immunization registry has powerful implications for health service delivery.
The predictive power of machine learning
Machine learning enables previously unachievable levels of precision and efficiency. A more precise supply chain is a more equitable supply chain. Machine learning is the science of teaching suites of algorithms (think recipes for how to perform an action, or analyze a specific set of information) to learn on their own and to identify meaningful patterns in data. Machine learning is technology that can observe and understand patterns in data. It can make predictions about the future, learning from mistakes and from success. Machine learning is the pinnacle of empowered data use.
The most efficient and advanced companies in the world employ machine learning to optimize core operations. Google reduced their energy costs 30 percent by injecting machine learning into how they manage energy; Amazon uses machine learning to anticipate demand for goods and stock products strategically; cardiologists use machine learning to read EKGs and radiologists use the technology to read MRIs.
The macro-eyes team is committed to using machine learning to improve health care. Using vaccine data from TImR, macro-eyes deployed core AI technology to predict vaccine use at a health facility level, for the weeks and months ahead. This granular visibility is unprecedented. If you know exactly how many vaccines will be consumed in the month ahead, you can correctly stock health facilities.
Early results from Tanzania are promising. When compared against the most accurate and currently used model for forecasting vaccine use, macro-eyes outperformed by more than 70 percent, reducing wastage and eliminating stockouts. Greater accuracy leads to lower costs, more efficient use of resources, and fewer missed opportunities for vaccination.
A trip to Tanzania
Last November, several members of the macro-eyes team visited Tanzania to share progress. Tanzania in November was hot for the macro-eyes contingent. Chief Medical Officer Dr. Frederic Bahnson and Chief Business Officer Drew Arenth both hail from the Pacific Northwest of the United States – a landscape famous for cool, drizzly rain. But the team had the BID Initiative by its side and the winds at their back. They were meeting directly with frontline health workers and sharing ground-breaking news with the Ministry of Health. The macro-eyes team had successfully developed the first cutting-edge machine learning technology to drive a predictive supply chain for vaccines.
BID and macro-eyes shared this exhilarating outcome with Dr. Dafrossa Lyimo, Tanzania’s Program Manager for Immunization and Vaccines Development. An innovator, unafraid to push the frontiers, she pressed hard on the need to put this technology to work immediately. Deploy, test, and iterate quickly. She emphasized the need to transition from the pilot phase and scale.
This work has potential to revolutionize vaccine delivery and break through the barriers restricting data use for the world’s most vulnerable populations. BID and macro-eyes are forging ahead to integrate this technology into every day workflow for the delivery of vaccines in Tanzania to improve the lives of millions.