Health Care, Food and Fairness

apple crates in warehouse

Photo by iStock/Smederevac

David Rea is algorithmically solving problems with data.

David Rea, an assistant professor of decision and technology analytics at Lehigh College of Business, works with hospitals and food distribution organizations to fix their real-world problems. His research integrates predictive and prescriptive analytics methodologies‚ largely focusing on operational problems in the delivery of basic needs services—food‚ shelter‚ health care‚ education.

“When it comes to feeding people‚ we have a mismatch between the availability of data and the reality on the ground‚” says Rea. Food insecurity is measured through an annual survey done in December. It asks people to reflect on the year and identify if they were worried about food or missed meals. The data is aggregated to a state level‚ and is released with about a two-year delay‚ according to Rea.

“This is in contrast with the hunger that people experience‚” says Rea. “Hunger is an immediate problem. There are emergency food organizations doing great work‚ but they’re doing it essentially blind, without the ability to understand if they are actually succeeding. They’re addressing hunger‚ not necessarily food insecurity. This makes it difficult to develop best practices and for those organizations to make the case for funding.”

The group of nonprofits Rea has been working with in Cincinnati is trying to create a real-time data system so that they can monitor hunger on a hyper-local level. Rea says they’ve been handling this by helping organizations integrate their data systems‚ and then using that data to drive real-time decision-making. “If there’s any good thing that came out of the pandemic‚” says Rea‚ “it’s the need and ability to integrate all these organizations.”

Rea is also using data to show how companies can algorithmically embed organizational justice—how an employee judges the behavior of the organization and the employee’s resulting attitude and behavior—inside their systems‚ decreasing turnover and increasing retention.

hospital/medical illustration
Photo by iStock/Elenabs

Rea’s research into emergency medicine also led him to a study where he and his colleagues examined physicians’ contracted scheduled clinical time across multiple emergency department locations.

“We were trying to look at when and where physicians prefer to work and how to reward or incentivize high contributing physicians in a fair manner‚” says Rea. “Using data, we were able to balance equity and equality with a single parameter and identify solutions that were efficient and fair, while reflecting what the emergency department wanted to incentivize.”

“A key point,” says Rea, “was the way equity was defined through a series of surveys that asked physicians what they thought should be included as a contribution. Inclusion in
the development process is critical to creating a system that is accepted.”

In another area of hospital department data research, Rea examined inter-hospital transport problems, a medical arena where there are multiple stakeholders—patients, crews, hospitals—with different goals.

“While hospital transport crews want balanced workloads, patients want quick response time for urgent cases,” explains Rea. “So, we are trying to create equal solutions with data for all stakeholders, but the system still needs to operate efficiently.” The researchers used the same method they did in the emergency room scheduling process. They created an objective to seek efficient solutions, while discouraging unequal and inequitable ones.

Rea says, “The key is trying to balance the treatment of your different stakeholders to find a solution that is acceptable to all.”

Listen to David Rea talk about his research.

Why it Matters

For managers, making large allocation decisions is difficult, even without considering the fairness of the results. As these decisions grow in complexity, they are increasingly being passed off to algorithms. Implementation of a decision algorithm, without careful consideration during their design, can create more problems than solutions. This research seeks to elucidate how an organization can transparently embed their values inside decision algorithms in a manner considered fair by all stakeholders.