<- Back to digest

European Spine Journal - 2026-10-16 - Journal Article

Accuracy of general-purpose large language models (LLMS) in interpreting adolescent idiopathic scoliosis radiographs: a comparative study with experienced spine surgeons.

Avcı İ, Kartal SB, Yavuz AY, Paksoy K, Yaman O

retrospective cohortLOE IIIn = 128 AIS radiographic case setsN/A

Topics

spinepediatrics
PMID: 42888262DOI: 10.1007/s00586-026-10443-zView on PubMed ->

Key Takeaway

General-purpose LLMs achieved 90.6–94.5% Cobb angle agreement within ±5° of surgeon consensus but showed substantially lower accuracy for Lenke classification, modifiers, and fusion-level landmarks, precluding independent clinical use.

Summary Depth

Choose how much analysis to show on this article page.

Summary

This study tested three general-purpose vision-capable LLMs (ChatGPT GPT-5.2 Instant, Gemini 3 Flash, Grok 4.1) against a three-surgeon consensus reference standard on 128 de-identified AIS radiographic case sets using a standardized Lenke-based prompt. Cobb angle estimation was acceptable across all models (MAE 2.4–3.1°; 90.6–94.5% within ±5°), but exact agreement for Lenke classification, lumbar modifier, sagittal modifier, neutral vertebra, and stable vertebra identification was substantially lower. The authors conclude these tools are unsuitable for independent measurement, surgical planning, or education without expert supervision.

Key Limitation

Surgeon consensus from only three surgeons at a single institution serves as the reference standard, introducing potential systematic bias if their collective measurement practice deviates from broader expert norms.

Original Abstract

PURPOSE

This study evaluates the accuracy of general-purpose, vision-capable large language models (LLMs) in interpreting adolescent idiopathic scoliosis (AIS) radiographs and compares their outputs with an expert surgeon consensus reference standard.

METHODS

This retrospective comparative diagnostic accuracy study analyzed 128 anonymized AIS radiographic case sets from 128 different patients, including standing coronal and sagittal radiographs and lateral bending films. ChatGPT GPT-5.2 Instant, Gemini 3 Flash, and Grok 4.1 were evaluated through their official web interfaces in January 2026 using the original standardized prompt based on the Lenke classification system. Model inputs were de-identified raw radiographs without measurement lines, Cobb values, vertebral labels, or Surgimap overlays. Three spine surgeons with more than 10 years of deformity surgery experience independently assessed all cases using Surgimap. Surgeon consensus was the reference standard. Cobb angle (CA) accuracy was assessed using mean absolute error and agreement within ± 5°; categorical variables were assessed by exact agreement with 95% confidence intervals.

RESULTS

ChatGPT achieved 94.5% agreement within ± 5° of surgeon consensus, with a mean absolute CA error of 2.4°. Gemini achieved 92.2% agreement with a mean absolute error of 2.7°, and Grok achieved 90.6% agreement with a mean absolute error of 3.1°. Exact agreement was substantially lower for Lenke classification, lumbar and sagittal modifiers, neutral vertebra, and stable vertebra identification. Inter-rater reliability among surgeons was high for CA and categorical parameters. No formal inferential ranking between LLMs was performed.

CONCLUSION

The tested general-purpose LLM versions showed limited reliability for comprehensive AIS radiographic interpretation. Their performance was substantially weaker for Lenke classification, modifiers, and fusion-level-related landmarks than for CA estimation. The findings do not support independent use of these systems for scoliosis measurement, education, or surgical planning. At present, these tools should be regarded as supervised documentation or teaching aids rather than measurement or decision-making systems.