A Mixed-Methods Evaluation of LLM-Based Chatbots for Menopause

TL;DR

This study evaluates publicly available LLM-based menopause chatbots using a mixed-methods framework focused on safety, consensus, objectivity, reproducibility, and explainability. It finds GPT-4o and Menopause Coach to be the most clinically aligned and safe, while other systems exhibit weaknesses in organization, sourcing, and empathetic guidance, particularly for mood and physical-symptom topics. The authors expose limitations of traditional metrics and document biases related to insurance status and race that affect guidance. They argue for ethically grounded, customized evaluation frameworks to improve reliability and safety in healthcare deployments of LLMs, and call for standardized methodologies to support real-world clinical use.

Abstract

The integration of Large Language Models (LLMs) into healthcare settings has gained significant attention, particularly for question-answering tasks. Given the high-stakes nature of healthcare, it is essential to ensure that LLM-generated content is accurate and reliable to prevent adverse outcomes. However, the development of robust evaluation metrics and methodologies remains a matter of much debate. We examine the performance of publicly available LLM-based chatbots for menopause-related queries, using a mixed-methods approach to evaluate safety, consensus, objectivity, reproducibility, and explainability. Our findings highlight the promise and limitations of traditional evaluation metrics for sensitive health topics. We propose the need for customized and ethically grounded evaluation frameworks to assess LLMs to advance safe and effective use in healthcare.