Gender-Neutral Large Language Models for Medical Applications: Reducing Bias in PubMed Abstracts

TL;DR

This study develops a bias mitigation pipeline for large language models in medical literature, neutralizing gendered pronouns in PubMed abstracts to promote fairness and reduce gender bias.

Contribution

Introduces MOBERT, a BERT-based model trained on gender-neutralized PubMed abstracts, to reduce gender bias in medical language models.

Findings

MOBERT achieved a 70% pronoun replacement rate.

Comparison showed MOBERT outperformed 1965BERT with only 4%.

Pronoun replacement accuracy linked to occupational term frequency.

Abstract

This paper presents a pipeline for mitigating gender bias in large language models (LLMs) used in medical literature by neutralizing gendered occupational pronouns. A dataset of 379,000 PubMed abstracts from 1965-1980 was processed to identify and modify pronouns tied to professions. We developed a BERT-based model, "Modern Occupational Bias Elimination with Refined Training," or "MOBERT," trained on these neutralized abstracts, and compared its performance with "1965BERT," trained on the original dataset. MOBERT achieved a 70% inclusive replacement rate, while 1965BERT reached only 4%. A further analysis of MOBERT revealed that pronoun replacement accuracy correlated with the frequency of occupational terms in the training data. We propose expanding the dataset and refining the pipeline to improve performance and ensure more equitable language modeling in medical applications.