2026 ASH®
Annual Meeting & Exposition December 12-15, 2026 ▪ New Orleans, Louisiana

Paper Details

Abstracts, authors, affiliations, and disclosures.

Saturday, December 12, 2026
05:30 p.m. - 07:30 p.m.
MCCNO - Hall F
Access, Representation, Population Health, Technology and Procedures, Imaging, Molecular Testing, Omics Technologies, Pathology, Bioinformatics, Blood Banking, Computational Biology, Emerging Technologies, Profiling, Treatment Considerations, Drug Development, Supportive Care, Therapy Sequence, Biological Therapies, Chimeric Antigen Receptor, (CAR)-T Cell Therapies, Women’s Health, Workforce, Adverse Events, Artificial Intelligence (AI), Computer Vision, Deep Learning, Generative AI, Machine Learning, Natural Language Processing, Biological Processes, Apoptosis, Genomics, Hematopoiesis, Molecular Biology, Clinical Practice (Health Services and Quality), Clinical Guidelines, Quality Improvement, Quality of Care, Systems-Based Hematology, Diseases, Bleeding and Clotting, Thromboembolism, Bone Marrow Failure Syndromes, Inherited Marrow Failure Syndromes, Paroxysmal Nocturnal Hemoglobinuria, Lymphoid Malignancies, Lymphoid Leukemias, Lymphomas, Plasma Cell Disorders, Myeloid Malignancies, Acute Myeloid Malignancies, Chronic Myeloid Malignancies, Metabolic Disorders, Iron Deficiency, Measurable Residual Disease, Non-Hodgkin Lymphoma, B Cell Lymphoma, Aggressive Lymphoma, Research, Fundamental Science, AML, APL, Translational Research, MDS, MPN, Clinical Research, Clinical Trials, Real-World Evidence, Study Population, Human, Adult, Pediatric
PAPER

2578 A Bayesian personalized reference interval framework predicts hemoglobin response to intravenous iron therapy

Cindy Zhang, Ya-Lin Chen2, Attila Jamilov3, Elvin Liu4, Swati Shree5, Barbara Lam Jr, MD6, Brody Harry Foy, DPhil7
1 University of Washington, Seattle, Washington, United States
2 University of Washington, Seattle, Washington, United States
3 University of Washington, Seattle, Washington, United States
4 University of Washington, Seattle, Washington, United States
5 University of Washington Medicine, Seattle, Washington, United States
6 Fred Hutchinson Cancer Center, Seattle, Washington, United States, University of Washington, Seattle, Washington, United States
7 University of Washington, Seattle, Washington, United States

Introduction:
Most routine laboratory markers, such as the blood count (CBC) have highly individualized homeostatic regulation, with values fluctuating around patient-specific setpoints. Despite this, clinical interpretation still relies primarily on static population reference intervals. However, modern availability of longitudinal electronic health record (EHR) data may allow for estimation of personalized reference intervals (perRIs) for common markers and analytes. We believe such perRIs may have value in both diagnostic and prognostic management of hematologic disorders.

Objective:
To develop a Bayesian inference framework for estimating personalized reference intervals from EHR data, and to evaluate whether deviation from hemoglobin perRIs is predictive of response to intravenous iron therapy.

Methods:
We analyzed outpatient laboratory test data from the University of Washington Medicine (UWM) EHR (n: 814,371 patients; 11,498,837 tests; 43 lab markers) over multiple decades. For each patient and marker, a Bayesian inference model was used to construct estimates of setpoints, and perRIs. For the targeted IV iron cohort, we used medication administration records to identify all UWM patients who received an intravenous iron infusion. We then considered their hemoglobin values within 60days pre-treatment, 60-180days post-treatment, and their historical hemoglobin perRIs from >1yr pre-treatment.

Results:
Our Bayesian inference framework generated highly individualized reference intervals that narrowed as patient history accumulated and were more predictive of future outpatient test results than static population reference intervals. Across diverse sets of markers (blood counts, metabolic panels, liver function tests, coagulopathic markers, etc.), setpoints and setpoint variability were consistently associated with increased mortality, as well as future morbidity signals. Across our iron infusion cohort (n: 492), personalized hemoglobin trajectories showed a consistent decline before infusion followed by recovery toward the individualized setpoint, with average recovery approaching the pre-disease setpoint approximately 3–4 months after treatment. The hemoglobin drop from setpoint to pre-treatment was more predictive of the post-treatment hemoglobin increase than the pre-treatment hemoglobin was alone (R² = 0.29 vs. 0.23, p = 0.001). In categorical analyses, patients with hemoglobins outside both the population and perRIs were more likely to achieve a >1 g/dL hemoglobin increase (“positive response”) than patients inside both intervals in (female: 76.2% vs. 35.8%, p < 0.001; male: 65.3% vs. 25.0%, p = 0.01). When considered separately, a pre-treatment hemoglobin drop below the personalized reference interval lower limit was more predictive of positive response than a pre-treatment hemoglobin below the population reference interval lower limit (females: 76.8% vs 42.1%, p<0.001; males: 64.0% vs 41.7%, p=0.03). These findings suggest that the hemoglobin drop relative to a patient’s own setpoint captures treatment-relevant physiology not fully represented by pre-treatment hemoglobin or population reference thresholds.

Conclusion:
Bayesian personalized reference intervals can transform longitudinal routine laboratory histories into individualized physiologic benchmarks. Patient-specific setpoints and marker variability were predictive of future mortality and morbidity across diverse laboratory markers. In patients receiving intravenous iron, personalized hemoglobin reference intervals were far more predictive of successful treatment response than population reference intervals. These findings support personalized laboratory setpoints as an emerging hematology decision-support tool.

Cindy Zhang: No relevant conflicts of interest to declare.