Findings discussed at today’s ADLM 2026 meeting
Anaheim, California, July 29, 2026 /PRNewswire/ — New research presented today at ADLM 2026 in Anaheim, California, suggests that machine learning (ML) may play a key role in improving the accuracy of standard blood tests used to diagnose rare tumors that form in or near the adrenal glands. This study also highlights the complexities of using ML models in laboratory medicine. As healthcare organizations seek to harness the power of ML, this study can help guide labs in rigorously validating ML algorithms by showing the types of pitfalls to watch out for.
Plasma-free metanephrines are the recommended first-line test to detect masses known as pheochromocytomas and paragangliomas (PPGLs), which cause overproduction of stress hormones in the body. If left untreated, PPGL can cause problems such as heart problems, headaches, and high blood pressure.
Metanephrine tests are effective at detecting people who have PPGL, but are less effective at excluding all people who do not have PPGL. That’s because mild elevations in metanephrines, a metabolite derived from stress hormones, are common in patients without these tumors, leading to false-positive results.
The researchers tested several ML algorithms to assess whether and how much accuracy could be improved by reducing the likelihood of these false positives. They analyzed data from 20,516 adults who underwent metanephrine testing at Samsung Medical Center between 2011 and 2024. Of the 19,797 patients tested who ultimately did not have PPGL, 25.2% showed elevated metanephrines, which can cause false-positive results.
“Our initial machine learning model suggested that combining plasma metanephrine results with structured clinical information from electronic health records may improve real-world discrimination,” said Se-eun Koo, one of the study’s co-authors and a clinical chemistry researcher in the Department of Laboratory Medicine and Genetics at Samsung Medical Center in Seoul, South Korea.
Specifically, Koo and co-author Soo-Youn Lee, Ph.D., found that while metanephrines have shown good reliability as a clinical marker, integrating ML to assess relevant clinical conditions (such as renal and urinary biomarkers, patient medications, and the presence of other diseases) appears to improve test performance and make the test superior.
But the story doesn’t end there. After submitting their initial findings to the Association for Diagnostic Laboratory Medicine (ADLM), Dr. Koo and Dr. Lee performed additional analysis to determine whether the ML algorithm was using “shortcuts” to learning that could introduce errors or bias. (Example: If ML consistently evaluates images of computer hackers wearing hoodie sweatshirts, it may conclude that anyone wearing a hoodie is a hacker.)
“After submitting the abstract, while testing the prototype app using the developed ML model, we realized that some of the apparent improvements may reflect patterns in the clinical workup rather than independent biochemical information,” Koo said. “Additional robustness analysis showed that much of the improvement was explained by shortcut learning from missing information.”
In other words, the ML model appeared to improve diagnostic accuracy, but it did so by learning which follow-up tests were ordered. This is what a clinician would do if they already suspected PPGL.
“This study shows that machine learning in laboratory medicine should be evaluated not only by performance metrics, but also by whether the model is learning the intended clinical signal,” Koo said. “That’s why external validation and careful auditing for learning shortcuts is so important.”
Koo will delve into this timely research project in a poster and oral presentation at ADLM 2026. The study itself is an interesting case study that illustrates the unique challenges researchers must face when evaluating artificial intelligence and ML.
“Our findings demonstrate both the potential and the pitfalls of applying machine learning to laboratory data,” Koo said. “The key message is not that machine learning is useless, but that routine care models need to be audited to ensure they are learning the intended clinical signals.”
Join Koo’s abstract session (details below) at ADLM 2026 to learn more about how she and her collaborators restructured their approach and developed new ML algorithms based on their discoveries about shortcut learning.
Session information
ADLM Registration for 2026 is free for media members. Reporters can register online here: https://xpressreg.net/register/adlm0726/media/landing.asp
Abstract B-091: Machine learning diagnostic performance of plasma metanephrines in PPGL: a large-scale real-world clinical cohort study Presented during:
science poster session
Wednesday, July 29th
9:30 a.m. to 5:00 p.m. (presenting authors present from 1:30 p.m. to 2:30 p.m.)
This session will be held in the Poster Hall on the Expo show floor at the Anaheim Convention Center in Anaheim, California.
About ADLM 2026
ADLM In 2026, the event will be held from July 26th to 30th in Anaheim, California, for five days filled with opportunities to learn about exciting science. Plenary sessions will explore groundbreaking diabetes research, the role of laboratory medicine in keeping astronauts healthy, the relationship between Alzheimer’s disease and Down syndrome, advances in cervical cancer screening, and the search for new cancer biomarkers using mass spectrometry.
At the ADLM 2026 Clinical Lab Expo, more than 700 exhibitors will fill the exhibit floor at the Anaheim Convention Center, showcasing the latest diagnostic technologies including, but not limited to, artificial intelligence, point-of-care, and automation.
About the Association for Diagnostic Medicine (ADLM)
Dedicated to achieving better health for all people through laboratory medicine, ADLM brings together more than 70,000 clinical laboratory professionals, physicians, research scientists, and business leaders from 110 countries around the world. Our community is comprised of individuals who are at the forefront of diverse subfields of laboratory medicine, including clinical chemistry, molecular diagnostics, mass spectrometry, clinical microbiology, and data science, and hold a variety of laboratory-related professional degrees, certifications, and qualifications. Since 1948, ADLM has championed advances in laboratory medicine by fostering scientific collaboration, knowledge sharing, and the development of innovative solutions that improve health outcomes. For more information, please visit: www.myadlm.org.
christine delong
ADLM
Director, Editorial and Media Relations
(p)202.835.8722
[email protected]
bill malone
ADLM
Senior Director, Strategic Communications
(p)202.835.8756
[email protected]
Source: Association for Diagnostic and Laboratory Medicine (ADLM)
