Building Personalized Reference Intervals for Routine Clinical Tests

Brody H Foy
University of Washington Medicine

Brody Foy, PhD (Presenter) University of Washington

Presenter Bio: Dr. Foy’s research is focused on utilizing large-scale clinical data to generate physiologic insights and improve patient outcomes, with particular focus on the hematologic, respiratory, and cardiac systems. He develops both machine learning and classical math modelling tools to better quantify patient physiology, develop novel biomarkers, improve forecasting, and enhance clinical workflows.

Relevant Financial Disclosures (within past 24 months, reported on Apr 16, 2026) No relevant financial relationship(s) to disclose.

Routine blood testing is a cornerstone practice in modern medicine. Common markers such as blood counts, metabolic panels, and liver function tests are used to inform effectively all areas of care, from family medicine to ICUs. However, while these markers reflect rich and dynamic physiologic systems, they are often collected, analyzed, and interpreted using crude and simple methods.

In this talk I will outline how modern AI and statistical methods can be used to enhance how we interpret and use routine testing. I will cover two major cutting-edge areas: 1) leveraging raw data streams from medical devices to enhance clinical inferences; and 2) leveraging longitudinal digital health records to create personalized reference intervals that are patient- and context-specific. I will show how careful analysis of these readily available, low-cost data streams can allow for creation of novel biomarkers, new prediction frameworks, and adaptive patient benchmarks, that can be readily applied across diverse populations. This in turn can enhance care, by allowing for earlier and more sensitive disease detection, while reducing unnecessary clinical workups. Our efforts highlight a novel approach to personalized medicine, built on widely and cheaply available routine laboratory data.