Artificial intelligence in endometriosis care: A comparative analysis of large language model and human specialist responses to endometriosis-related queries

Laurin Burla, Julian Matthias Metzler, Dimitrios Rafail Kalaitzopoulos, S. Kamm, Simone Kamm, Mark Ormos, Daniel Passweg, Sabrina Schraag, Eleftherios Pierre Samartzis, Nicolas Samartzis, Isabell Witzel, Patrick Imesch
European journal of obstetrics, gynecology, and reproductive biology · 2025 · vol. 313 , pp. 114625 · doi:10.1016/j.ejogrb.2025.114625 · PMID:40829501 · W4412988157
article OA: hybrid CC0
🔓 Open OA copy View on OpenAlex View on PubMed View at publisher
⚙ AI-generated summary by claude@2026-06, 2026-06-07 ⓘ

This study found that large language models were not inferior to human specialists in answering endometriosis-related queries, with no substantial differences observed in medical content despite reviewers being able to significantly distinguish between AI and human responses.

One-sentence paraphrase of the abstract; not a substitute for reading it. No clinical advice. How this works

Abstract

INTRODUCTION: Many women with endometriosis turn to digital sources for information. Meanwhile, large language models (LLMs) appear capable of offering health advice. This study aimed to evaluate the potential in answering endometriosis-related queries. METHOD: This comparative study used 150 anonymized endometriosis Q&As from online forums (2021-2023), answered by human specialists. Another 150 responses were generated using an LLM (ChatGPT-4o). Eight expert reviewers, split into two groups, blindly evaluated either the human or artificial intelligence (AI) response for each question. The primary endpoint was whether responses could be correctly identified as human or AI. Secondary endpoints included incorrect information, harm, and suitability for patient communication. RESULT: = 246.162, p < 0.001, slight interrater agreement). Most responses were accurate (84.8 %), harmless (87.2 %), and patient-suitable (73.8 %). There were no significant differences regarding incorrect information (p = 0.308), harm likelihood (p = 0.944), harm extent (p = 0.892), medical consensus alignment (p = 0.235), or suitability for patients (p = 0.544). CONCLUSION: This study found that AI was not inferior to human specialists in answering endometriosis-related queries. While reviewers were able to distinguish AI- from human-generated responses significantly, interrater agreement was only slight. No substantial differences were observed in medical content. As AI continues to evolve, patients increasingly turn to it for medical guidance, highlighting the need for greater specialization in endometriosis care. Future research should further investigate the risks, benefits, and patient acceptance of AI in clinical practice.

My notes (saved in your browser only)

Condition tags

mesh:D004715endometriosis

MeSH descriptors

Artificial Intelligence Artificial Intelligence Artificial Intelligence Artificial Intelligence Artificial Intelligence Artificial Intelligence Artificial Intelligence Artificial Intelligence Artificial Intelligence Artificial Intelligence Endometriosis Endometriosis Endometriosis Endometriosis Endometriosis Endometriosis Endometriosis Endometriosis Endometriosis Endometriosis

Citation neighborhood

Papers in the corpus that this work cites (lower rings, blue) and that cite this one (upper rings, green). Dot size scales with the paper's in-corpus citation count — bigger dot = more influential within the endo/adeno field. Click a dot to open that paper. [ expand to 2 hops ] — adds papers reached through this work's immediate citers/citees. Heavier; up to 60 extra dots.

References (33)

Source provenance

europepmc
last seen: 2026-06-04T01:30:01.192114+00:00
openalex
last seen: 2026-06-04T00:00:01.174412+00:00
pubmed
last seen: 2026-06-02T00:31:12.050562+00:00
License: CC0 · commercial use OK