Elizabeth Bowen, MD, clinical assistant professor in Endocrinology; Nitin Pagedar, MD, MPH, professor of Otolaryngology; and Zachary Urdang, MD, PhD, postdoctoral research scholar, were awarded a $15,000 pilot grant from the University of Iowa College of Engineering for their project “Using Machine Learning to Predict Normocalcemic Primary Hyperparathyroidism.” Funding from this grant will be used to develop a machine learning algorithm to help identify patients who may have normocalcemic primary hyperparathyroidism (nPHPT).

nPHPT is a condition that presents with elevated parathyroid hormone and normal serum calcium levels. It is caused by abnormal growth and functioning of one or more parathyroid glands.

“As patients with normocalcemic primary hyperparathyroidism present with normal serum calcium levels, their biochemistry is essentially identical to those with hyperparathyroidism as a physiologic response to low calcium intake or absorption, also known as secondary hyperparathyroidism,” Bowen said. For this reason, nPHPT is difficult to diagnose.

Together with Yanan Liu, a data scientist from the College of Engineering, Bowen, Pagedar, and Urdang plan to develop a machine learning algorithm to improve detection of this disease, and hopefully help physicians distinguish nPHPT from secondary hyperparathyroidism. Using the TriNetX database, a source of de-identified data from over 100 million patients worldwide, common variables such as serum calcium and phosphorus, vitamin D, creatinine, systolic and diastolic blood pressure, age, BMI, weight, race, history of osteoporosis medications, and ICD-10 codes indicating a history of kidney stones, osteoporosis, osteopenia and/or fractures will be analyzed. Of these variables, those that are most predictive for diagnosing or excluding nPHPT will be included in the algorithm. The algorithm will then be evaluated for its ability to predict or exclude nPHPT. Once pilot data is available, the team will apply for a larger grant to improve upon the algorithm to study outcomes related to its application.

Bowen’s and Pagedar’s expertise in diagnosing and treating primary hyperparathyroidism, Urdang’s role as consultant for TriNetX and extensive experience with machine learning as a clinical tool, and the team’s collaboration with the Iowa Initiative for Artificial Intelligence combine to introduce potential for comprehensive and cutting-edge research and enhanced clinical care.