Posted on October 18, 2026 by Wendy Frost
How does an overtaxed healthcare system deal with increasing demands to treat chronic diseases coupled with shortages in medical professionals?
Katherine Adams
Beyond medical interventions, research in the Carlos Alvarez College of Business has shown that optimizing healthcare delivery can significantly improve health outcomes.
Katherine Adams, a postdoctoral fellow in operations and analytics in the Alvarez College of Business, and her co-authors have developed a prescriptive analytics model that reduced fasting blood glucose levels in diabetics by 25%.
Her paper, “Planning a Community Approach to Diabetes Care in Low-and Middle-Income Countries Using Optimization,” was accepted in the top-ranked journal Operations Research.
“Machine learning is typically used for predictive analytics, forecasting something that will happen in the future,” said Adams. “People don’t know enough, in my opinion, about prescriptive analytics, like optimization. It’s about trying to make the best decision you can with the information and resources you possess.”
The field of operations research got its start during World War II when a structured approach was needed to solve large-scale resource allocation problems. It uses mathematical modeling and statistical analysis to solve complex decisionmaking problems. It is applied across almost every industry including manufacturing, logistics, transportation and healthcare.
For Adams’ paper, she collaborated with NanoHealth, based in India, to address a problem they had regarding the best way to allocate their community health workers to identify and manage diabetes patients.
Using patient data from community health workers’ visits, she developed an optimization framework to maximize glycemic control for that community that accounts for patients’ disease progression and enrollment decisions.
Using data points like initial and current fasting blood glucose levels, blood pressure, body mass index, number of visits and frequency of visits, Adams’ model can help workers decide which patient to visit to better control blood glucose levels for as many patients as possible.
The model also incorporated each patient’s motivational state based on their health changes as well as how motivated they were to stay in the program.
“One of the findings is that modeling the patient motivational state does help us allocate resources more efficiently as it allows us to anticipate drop out decisions,” she said. “In addition to personalizing the visiting decision, the workers can consider the tradeoffs when determining the balance between new patient enrollment and management visits.”
By testing this process using NanoHealth data as well as simulated data, her model matched the performance of existing approaches while using up to 50% fewer worker visits.
“A lot of operations research tends to focus on maximizing profit or reducing cost,” said Adams. “We have the knowledge, medicines and treatment options, but why are people still suffering when it is a manageable condition? That is why I’m passionate about this work.”
— Wendy Frost
Share this story:
