Research
PANKUSH KALGOTRA Associate Professor Department of Business Analytics and Information Systems
In 2023, Kalgotra published research using over 45,000 medical histories to predict early onset cancer.
AI: The Modern Oracle L ong ago, the ancient Greeks tried a lot of things to predict the future. They consulted oracles, cast bones,
population under the age of 50 years. They represented the relationships between
with complex analysis quick and comprehensible inside the virtual environment. Equipped with headsets and immersed in a space with Generative AI capabilities, users see a visualization of the data, specifically, network of diseases. With voice commands, AI can be ‘asked’ questions about the data and directed to refine response or bring in outside sources to contribute new insights. In today’s world, where AI takes care of the technical grind, the real skill is knowing what to ask and why it matters—because the future belongs to those who ask the right questions. Ask a question, request a prediction and you’ll get an answer. A lot easier than looking at entrails. HM
different diseases as a matrix and recorded how often various diseases co-occur. They created matrices for patients with colorectal cancer, and those without, and calculated the specific variables that quantify disease interactions. The researchers could then measure how a patient’s profile matched other known colorectal patients and train the AI model to predict the likelihood of the disease. Their model enabled them to identify individuals with no family history at potentially higher risk, even before the onset of symptoms. This approach is a cost effective and scalable triage tool that can proactively identify high-risk individuals at a younger age which in turn allows early screening, intervention and improved care. And it’s not just the healthcare industry that can benefit from Kalgotra’s predictive modeling. Any industry with data sets can utilize AI (and most recently Generative AI) to uncover information that might be hidden from human eyes and predict outcomes based on that data. Kalgotra’s next effort is the design of a Virtual Reality user interface that can make the interaction
mapped celestial bodies, analyzed the flight of birds and examined the entrails of sacrificed animals for signs and omens. Over time, we’ve looked to palm readers and tarot cards and tea leaves. Now, perhaps a bit more rational, we tend to trust scientists, scholars, and most recently, AI. AI, trust? Maybe if that trust is backed up by solid unbiased databases and rigorously trained generative AI. Harbert Associate Professor Pankush Kalgotra used AI to explore volumes of data, the sheer size of which defies human analysis. AI models are built on neural networks—computational models patterned after the structure of the human brain. This structure allows the model to learn and understand the subtle patterns within large data sets. In 2023, Kalgotra, with fellow researchers, Dr. Ramesh Sharda of Oklahoma State and Seattle gastroenterologist Dr. Sravanthi Parasa, used their AI model to digest over 45,000 medical records, looking to predict colorectal cancer in the
ABOUT THE RESEARCH: Kalgotra, P., Sharda, R., & Parasa, S. (2023). Quantifying disease-interactions through co-occurrence matrices to predict early onset colorectal cancer. Decision support systems, 168, 113929. https://doi.org/10.1016/j.dss.2023.113929
10 HARBERT.AUBURN.EDU
Powered by FlippingBook