A new study published in the Journal of the American Medical Informatics Association (JAMIA) showcases how modern AI methods are advancing clinical and translational research in Neurofibromatosis type 1 (NF1). Conducted by Levi Kaster, BS, Ethan Hillis, MS, Inez Y. Oh, PhD, Elizabeth C. Cordell, MD, Randi E. Foraker, PhD, MA, Albert M. Lai, PhD, Stephanie M. Morris, PhD, David H. Gutmann, MD, PhD, Philip R. O. Payne, PhD, and Aditi Gupta, PhD, the research aims to extract valuable insights from unstructured clinical data to improve NF1 patient care.
NF1 is a complex genetic disorder affecting various organ systems. The study developed pipelines using rule-based natural language processing (NLP) and large language models (LLMs) to identify NF1-related phenotypes in clinical notes. The results revealed that while rule-based NLP showed higher precision after refinement, LLMs had better generalizability without needing adjustments.
This research underscores the potential of AI-driven tools to enhance the portability of clinical data extraction methods, ultimately aiming to improve disease management for NF1 patients.