A Large Language Model Outperforms Other Computational Approaches to the High-Throughput Phenotyping of Physician Notes

TL;DR

This paper demonstrates that a Large Language Model, specifically GPT-4, outperforms traditional NLP and hybrid methods in automating the extraction of phenotypic information from physician notes, advancing precision medicine.

Contribution

The study introduces and evaluates a Large Language Model approach for high-throughput phenotyping, showing its superiority over existing computational methods.

Findings

GPT-4 achieved higher accuracy in phenotyping tasks.

Large Language Models outperform traditional NLP methods.

The approach enhances automation in processing electronic health records.

Abstract

High-throughput phenotyping, the automated mapping of patient signs and symptoms to standardized ontology concepts, is essential to gaining value from electronic health records (EHR) in the support of precision medicine. Despite technological advances, high-throughput phenotyping remains a challenge. This study compares three computational approaches to high-throughput phenotyping: a Large Language Model (LLM) incorporating generative AI, a Natural Language Processing (NLP) approach utilizing deep learning for span categorization, and a hybrid approach combining word vectors with machine learning. The approach that implemented GPT-4 (a Large Language Model) demonstrated superior performance, suggesting that Large Language Models are poised to be the preferred method for high-throughput phenotyping of physician notes.