By Shannon Turgeon
Photography by Joshua Franzos
As an associate professor of biostatistics and health data science at the University of Pittsburgh School of Public Health, Lu Tang acknowledges that he finds fun in looking at mathematical equations.
However, his research and collaboration across Pitt’s schools of the health sciences have motivated him to dream bigger: to develop new methodological frameworks that can be applied to real-world data, improving clinical decision making and patient outcomes.
“It's more rewarding to really see what statistics can bring to life—what impacts they could have on people, and what knowledge we could extract based on these models being applied to a concrete, specific problem of interest,” said Tang.
On Friday, Dec. 12, Tang will present “Advancing Precision Health Through Integrative Analysis of Real-World Data” as part of the 2025 Senior Vice Chancellor’s Research Seminar Series. (Join the lecture here.)
Tang’s research focuses on how to integrate multisource, real-world health data. He works with electronic health records from hospital data networks and insurance claims from multistate Medicaid networks. He then uses machine learning and statistical methods to combine and analyze this data.
His specific frameworks are unique because they allow data from multiple hospitals to be combined without sharing sensitive patient information. They also enable the use of statistical methods, such as causal inference, a fundamental methodology in statistics that is used to understand cause and effect. This could lead to individualized treatment recommendations for patients without compromising their personal data.
Each hospital analyzes its own data privately and learns patterns related to patient type, treatment response and causal relationships. These insights—not patient records—are then shared between hospitals. The combined information allows for specific recommendations that are based on nonidentifying characteristics, such as “elderly patients” or “patients with low blood pressure.”
An example of this can be seen in Tang's collaboration with the School of Medicine's Department of Critical Care Medicine. He used electronic health records from sepsis patients across UPMC hospitals to determine the best time to administer treatments like fluids and vasopressors—and for which patients—to improve survival rates.
“If we apply a one-size-fits-all approach to everybody, patients will not get what's the best for them—they're going to get what's the best for the entire population. Every patient is unique, and patients may need specific, individualized plans to help them recover so that they can get the best outcome possible,” he said.
Tang keeps the principles of safety, interpretability and fairness at the forefront of his work. His goal is to develop trustworthy artificial intelligence methods that can be used in day-to-day clinical practice.
“In the health care and health policy settings, we’re dealing with more complicated data, and high stakes problems that could lead to both death and economical losses. That really drives me to think from these three perspectives when developing new methodologies,” he said.
In the future, Tang hopes to see his research have a direct and far-reaching impact in the medical field.
“One day, I hope to see that the approaches we developed could actually help save someone, or avert a life-threatening condition,” he said.
“We are training the next generation of our students, biostatisticians and health data scientists to be prepared to translate their work into impactful, real-world solutions.”