Better data science, better societies
Ram Bala talks about the “algorithmic juice” that has enabled GetUsPPE to deliver two million PPE units across United States during the COVID-19 crisis, and why he believes data science has a vital role in solving society’s biggest challenges
In many instances, COVID-19 has shone a bright light on the goodness of people. With our backs against the wall, we have pulled together in different ways to support each other and navigate the colossal challenges thrown up by a pandemic.
One such challenge has been preparing our healthcare workers for the essential duties they continue to perform day-in, day-out, to stem this coronavirus. The supply of personal protective equipment, or PPE, has been a critical problem throughout the pandemic; as early as March 3rd, the World Health Organization was calling on industry and governments to increase PPE manufacturing by 40%.
Yet while PPE supply remains a complex issue even now, especially in the United States where a fresh surge of COVID-19 cases is putting further strain on healthcare systems, extraordinary stories of collaboration have emerged. One example is the rapid rise of GetUsPPE (GetUsPPE.org).
GetUsPPE is a grassroots coalition of medical professionals, scientists, programmers and citizens, formed in the early stages of the pandemic to confront head on the PPE predicament in the U.S.. Initially created to solve an acute but short-term issue, GetUsPPE has grown into a considerable force for good, working alongside more than 12,000 hospitals and connecting PPE donors with those who need it most. It is a heartening tale of volunteers battling against the silent enemy of COVID-19.
Matching PPE supply with demand, however, is by no means a straightforward task. GetUsPPE has relied upon tailor-made technology to help it deliver more than two million PPE units nationally during the crisis. Ram Bala, the technological driving force for GetUsPPE and now a member of its advisory board, picks up the story.
“I got thrown into a Slack feed in late March and it was chaos!” reflects Bala with a chuckle. “You could see hundreds of people swarming on this problem, people who had never met each other and whose only knowledge of each other came through these online channels. They were all trying to solve the problem of supplying PPE to healthcare workers.”
Bala’s background and expertise meant he was the perfect person at the perfect time for GetUsPPE. Since completing a PhD in optimisation models and analytics related to market places and supply chains, Bala has combined an enviable academic career — he is currently as Associate Professor at Santa Clara’s Leavey School of Business — with numerous ventures in supply chain and logistics.
He has made a career out of applying data science to drive new efficiencies in these fields, so it’s no surprise that the supply and demand issues faced by GetUsPPE piqued Bala’s interest. He talks about how supply “mismatching” is a common problem in similar economic models.
“In any distribution network, whether it’s PPE or not, the biggest challenges are the informational and logistical gaps which cause mismatches in supply and demand. Some people have too much, and some people have too little,” Bala explains to Tech For Good.
“At the beginning of the pandemic, I read newspaper articles about how there was going to be a shortage of PPE. Then a lot of the leaders in this field were writing to say that, yes, there was inadequate supply, but it was also about the fact that even with the supply they had, they didn’t have the right ability to match that supply with the demand.”
After posting on LinkedIn in March to offer his professional expertise, Bala was put in touch with GetUsPPE via a friend of a friend. GetUsPPE — co-founded by Megan Ranney, the emergency physician, Brown University professor and high-profile voice on the topic of PPE shortage — was then kick-starting its mission but had a “primitive” database, according to Bala.
“You had two Google Sheets, one of supply and one of demand. Donors would say ‘this is how much we have’ and hospitals would say ‘this is what we need’. And they were just figuring out what the best way was to move the PPE around. The easiest way to do it is to say ‘hey, here’s this hospital close by’, but what if that hospital only wants 200 masks, and they have 400 masks? Why should they leave these masks there? They should be sending them somewhere else.”
Using test data, Bala and his volunteer team got to work on a unique optimisation algorithm to solve GetUsPPE’s problem. In a short space of time, they turned it into a cloud-based service that GetUsPPE’s team members could benefit from straight away. The system generated an output that told them exactly how and where to move the PPE.
Bala labelled the work “Project Stanley”, which quickly became the “algorithmic juice” that supported GetUsPPE in its transformation from a social media hashtag into a successful non-profit enterprise operating at national scale — and having a significant impact in the fight against COVID-19.
As the movement progressed quickly, the coders had access to more data — critical, in the eyes of Bala — and have been able to refine the algorithm to deal with additional complexities.
“One challenge is geography, the United States is just a very big country,” states Bala. “Sometimes you have to solve problems where they may not be local — if a large donor comes in and has a million items, how do you move those?
“Then there is the entire bioethics area — how do you prioritise PPE based on what kind of population the hospital serves? Or what the vulnerability of a specific area is? The model is now able to take into account some of these priorities, figuring exactly what the cost-benefit is in terms of where to move things effectively while maximising impact.
“Part of the problem, as with any application, is getting the right data. GetUsPPE probably at this point has the best supply and data in the United States, certainly demand data, in terms of what hospitals actually need.”
Bala speaks with great pride at the success of Project Stanley and is full of admiration for the work his colleagues at GetUsPPE have done in recent months. The experience has had such an impact on him personally that Bala’s new aim is to take Project Stanley and apply its technology and talent to other non-profit and humanitarian organisations.
Project Stanley has itself been incorporated as a non-profit in California, with a 501(c)(3) status pending, and Bala is adamant its cutting-edge approach to data science can help others tackle their supply chain, logistics and marketplace issues.
“Personally I was really surprised how successful it was,” he admits. “I didn’t know whether it was going to get used or not, because I’ve been in the for-profit space as well, and sometimes you build things and adoption rates are low.
“With GetUsPPE, it was wonderful. They said ‘this is great, it’s going to work for us’. And they made it part of their workflow, and suddenly we were just seeing this incredible use of the algorithm. So we kept enhancing it, and at that point the discussion in my own team became around the idea that this was something that could have wider applicability.”
One use case Bala is particularly excited about is in food supply chains. Food and agriculture are normally big business in California — it is the leading U.S. state for cash farm receipts — but it is another sector that has been hit hard by COVID-19.
Through colleagues at Santa Clara University, Bala became acquainted with a member of San Jose’s newly-formed Resilience Council and learned that many of the city’s food banks were getting overwhelmed, with the food banks themselves finding it hard to predict how many locals would visit at any given time.
At the same time, many of California’s farms were seeing increasing amounts of food waste because demand in their own supply chains had fallen drastically. Bala saw a similar opportunity to the one Project Stanley grasped with GetUsPPE.
“The idea is to eventually come up with a predictive dashboard that will marry the demand for food with access points, so trying to measure where the gaps are and coming up with better metrics for managing that, and the dashboard will be open to anybody” he explains.
“I see a revolution coming. There are all these teams who are data and tech-savvy, who are all looking at these datasets and asking if they can connect the dots and make more meaningful sense of what is happening? It’s an exciting time”
“I think San Jose specifically needs it because in this pandemic, the county has told them to help the food banks. But we’re not just thinking about this as a local thing. If we have a model that works for the Bay Area, I believe this model can be expanded across the United States. We should have a national map of this which can be drilled down into.”
Project Stanley’s ambitions don’t stop there. Bala is engaged in a number of conversations with other non-profits in agriculture, education and healthcare, and in partnership with Manifolds Lab LLC, the Starling Lab at UT, Dallas and GetUsPPE, Project Stanley has submitted a grant proposal to IARPA for Project LOTUS (Logistics Optimized Transparently for Urgent Supply), an ambitious initiative focused on next-generation emergency logistics.
Bala even says he has been involved in “active discussions” around how the vast amounts of data at play could help combat some of society’s biggest issues.
“I see a revolution coming. There are all these teams who are data and tech-savvy, who are all looking at these datasets and asking if they can connect the dots and make more meaningful sense of what is happening? It’s an exciting time,” he concludes.
“For me, what’s been particularly heartening is the fact there has been so much investment of time from so many people without much thought about how they’re going to make money out of it. Altruistic seems too strong a word, but people are more concerned with solving these problems than about what they’re going to make out of the end of it. That’s great — particularly in times of crisis, we need that. If we attach dollar value to everything, it’s hard to solve real problems of this kind.
“I just feel at a fundamental level, as the wealthiest country in the world, these are problems we should be able to solve.”