Excellence Unveiled: Michael Kosorok
The W.R. Kenan Jr. Distinguished Professor of Biostatistics talks about how artificial intelligence can be used to advance health research and help solve public health challenges.
Press play above to listen
Each month, WCHL will feature stories highlighting Carolina excellence from outstanding Tar Heels to share why they’re passionate about their work and shine a light on their important contributions to the community, state and beyond.
Tune in to Excellence Unveiled during morning, noon and evening drive times to hear their stories and find out what ignites their passion for their work. Excellence Unveiled is an exclusive program on WCHL, sponsored by UNC-Chapel Hill. You can listen to WCHL at 97.9 FM or 1360 AM.
Read a transcript of the segment below:
Read a transcript of the segment below:
Chancellor Lee H. Roberts: Welcome to Excellence Unveiled, sponsored by the University of North Carolina at Chapel Hill. I’m Chancellor Lee Roberts, and it’s my pleasure to shine a light on our talented Carolina community and its outstanding work. This month, we feature Michael Kosorok, W.R. Kenan Jr. Distinguished Professor of Biostatistics, to talk about how AI can be used to advance health research and help solve public health challenges.
Michael Kosorok: So one thing that I think is unique overall about UNC is the collaborative culture. And the thing that completely blew me away was how much people like to work together in different disciplines. We have very, very deep expertise here. Artificial intelligence is such a big area that everybody thinks of all sorts of things when you hear the word. Our focus is on something called generative AI and large language models and neural nets, as well as reinforcement learning. This just means that we don’t do just the sort of standard machine-learning type analysis. We’re really focusing on some of the edgier types of AI tools. Our mission is to help our university become leaders in cutting edge artificial intelligence to improve human health. We are working on both the applications of these. We want to develop tools that can be used to improve health. We also want to develop methods so that we can be on the cutting edge.
Well, there are actually several exciting projects. I can tell you one that is completed mostly, and this is one that involves work with Jeffrey Stringer and his group. The goal of this was to develop a tool that can be used in very under-resourced locations to accurately predict gestational age without the usual resources. So normally when women are pregnant and looking forward to having a child, there’s a physician, there’s an ultrasound technician and ultrasound equipment that’s used to see how the baby is doing and assess gestational age. In under-resourced locations, you may have none of those things available. And so we had to develop something that can be used in these locations successfully. My lab, and especially a student of mine at the time, Ben Pokaprakarn, helped contribute to the AI development itself. The tool, our first publication, showed that this was very, very accurate — in fact, more accurate than an expert.
One other example is in the area of lower back pain. This is a collaboration with 13 institutions around the country. UNC-Chapel Hill is the coordinating location. I helped with the reinforcement and the design of the study, and a student of mine did the majority of the work. In this study, we’re looking at how can we help physicians know when and how to treat patients with lower back pain. This is a chronic problem, and people feel it’s one of the main causes of the opioid epidemic because it’s so chronically painful. We develop a special type of trial design, and our goal is to be able to use reinforcement learning, which is a type of AI, to link the different biomarkers that a patient has, different measurements, where they’re at in their health, to be able to make sure that the first treatment is the best possible and that the next treatment is the best adaptation based on how the first treatment went. Over 900 patients were entered into the study, and so that analysis we hope to complete this year.
Chancellor Lee H. Roberts: This has been Excellence Unveiled, proudly presented by the University of North Carolina at Chapel Hill.
