RareDAI follows a clinician’s chain-of-thought, allowing for reproducibility based on a series of standardized questions
PHILADELPHIA, June 24, 2026 /PRNewswire/ — Researchers from Children’s Hospital of Philadelphia (CHOP) have fine-tuned large language models (LLMs) to analyze clinical data, aiding medical teams in determining which available genetic tests are the appropriate options to help diagnosis rare diseases. This approach could help save time and increase the reproducibility of clinical decision making. The findings were recently published in the journal NPJ Digital Medicine.
Many professional societies offer clinical guidelines for genetic testing. However, with thousands of known genetic disorders and many more being discovered, it can be challenging to select between gene panels and whole exome or whole genome sequencing. Additionally, the guidelines are meant to cover a wide variety of possible diseases, often leading to some ambiguity.
“The implementation of the guideline within a particular healthcare system has always been a huge challenge, especially for those guidelines that don’t have a highly specific protocol to implement,” said senior study author Kai Wang, PhD, Professor of Pathology and Laboratory Medicine at CHOP. “Without a concrete threshold, it’s left up to human interpretation, which can lead to variability.”
In this study, Wang and his colleagues trained AI-driven models to interpret why clinicians choose certain genetic testing modalities based on seven key questions, addressing important concerns like whether the condition is a suspected congenital abnormality or developmental disorder, whether there is a family history of genetic disease, or whether there are cost or accessibility concerns.
This led to the development of RareDAI, an integrative approach that addresses the challenge of following a clinician’s chain-of-thought and can aid others in following a similar workflow for the patients they encounter. This tool essentially shows the work of how it led to a particular recommendation so that a clinician can understand the thought process before choosing to follow the recommendation. The study found that RareDAI outperformed traditional fine-tuning and base large language models by up to 20% across all metrics like accuracy and precision.
“Our paper demonstrates that AI can indeed have high consistency with human decision-making, but at the same time make the decision-making process interpretable and kind of reproducible, allowing for decision-making to be documented and examined at a future date,” Wang said.
This study was supported by National Institutes of Health grants HG012655, HG013031, HD111688 and the Children’s Hospital of Philadelphia Research Institute. The study was conducted in collaboration with Dr. Chunhua Weng at Columbia University and Dr. Wendy K. Chung at Boston Children’s Hospital, both of whom are multiple principal investigators on the NIH HG013031 grant supporting this line of research.
Nguyen et al, “Interpretable fine-tuned large language models facilitate making genetic test decisions for rare diseases.” NPJ Digit Med. Online May 19, 2026. DOI: 10.1038/s41746-026-02733-z.
About Children’s Hospital of Philadelphia:
A non-profit, charitable organization, Children’s Hospital of Philadelphia was founded in 1855 as the nation’s first pediatric hospital. Through its long-standing commitment to providing exceptional patient care, training new generations of pediatric healthcare professionals, and pioneering major research initiatives, the hospital has fostered many discoveries that have benefited children worldwide. Its pediatric research program is among the largest in the country. The institution has a well-established history of providing advanced pediatric care close to home through its CHOP Care Network, which includes more than 50 primary care practices, specialty care and surgical centers, urgent care centers, and community hospital alliances throughout Pennsylvania and New Jersey. CHOP also operates the Middleman Family Pavilion and its dedicated pediatric emergency department in King of Prussia, the Behavioral Health and Crisis Center (including a 24/7 Crisis Response Center) and the Center for Advanced Behavioral Healthcare, a mental health outpatient facility. Its unique family-centered care and public service programs have brought Children’s Hospital of Philadelphia recognition as a leading advocate for children and adolescents. For more information, visit https://www.chop.edu.
Contact: Ben Leach
Children’s Hospital of Philadelphia
(609) 634-7906
[email protected]
View original content to download multimedia:https://www.prnewswire.com/news-releases/childrens-hospital-of-philadelphia-researchers-develop-ai-driven-tool-to-aid-in-selecting-genetic-tests-for-diagnosis-of-rare-diseases-302809640.html
SOURCE Children’s Hospital of Philadelphia
