Development and multi-center validation of a pre-trained language model for predicting neonatal morbidities

Presentation Time: 04:00 PM - 04:15 PM

Abstract Keywords: Pediatrics, Natural Language Processing, Clinical Decision Support, Large Language Models (LLMs), Critical Care
Primary Track: Applications
Programmatic Theme: Clinical Informatics

We present work in developing, training, and validating NeonatalBERT, a pre-trained language model to automatically predict neonatal diseases at birth from unstructured clinical notes based on a large dataset with over 30,000 newborns. We perform both internal and external validation on a comprehensive list of neonatal morbidities and demonstrate strong performance across hospitals and patient populations. NeonatalBERT has a great degree of flexibility and paves the way for various future applications in neonatal care.

Speaker(s):
Feng Xie


Author(s):
Feng Xie; Philip Chung, MD, MS; Jonathan Reiss, MD - Stanford University; Erico Tjoa, PhD - Stanford University; Thanaphong Phongpreecha, PhD - Stanford University; William Haberkorn, MS - Stanford University; Dipro Chakraborty, MS - Stanford University; Alan Chang, PhD - Stanford University; Tomin James, PhD - Stanford University; Yeasul Kim, MD - Stanford University; Samson Mataraso; Ivana Maric, PhD - Stanford University; Sayane Shome, PhD - Stanford University; Momsen Reincke, MD - Stanford University; Gary Shaw, DrPH - Stanford University; David Stevenson, MD; Nima Aghaeepour - Stanford University;