PulseTrendingJournal ClubResearchersJournalsExplore
Instagram
HomeTrendingJournal ClubExplore
Synapse
⌘+K
Synapse
July 26, 2026MedicineOpen Access

Performance of DeepSeek V3 and ChatGPT-4o in answering esophageal cancer-related questions

View Full Paper
Ask AI
Bookmark
Share

Authors

QYQian YangJYJing YiAGAnran Gong

Discussion

Loading...

Member takes

Overview

Randomized trial assesses accuracy of AI models in answering esophageal cancer-related questions, highlighting potential knowledge gaps.

Key Points

  • This study aims to evaluate the accuracy of DeepSeek V3 and ChatGPT-4o in responding to health-related knowledge questions about esophageal cancer.
  • Fifty-two esophageal cancer-related questions were classified into themes and input into DeepSeek V3 and ChatGPT-4o.
  • Responses were evaluated by two experienced gastroenterologists for accuracy and temporal stability.
  • Scoring was conducted on a scale of 1 to 4 to assess responses across different categories.
  • DeepSeek V3 scored 4 (3–4) across most categories; ChatGPT-4o scored 4 (3–4) for most but 3 (2–4) in diagnosis.
  • No statistically significant differences were found between the scores of the two models (P > .05).
  • Both models showed some inconsistency; DeepSeek V3 on 2 questions and ChatGPT-4o on 1 question.

Cite This Study

Yang et al. (2026) studied this question.

synapsesocial.com/papers/6a65a91fd3aea3239cd7911fhttps://doi.org/10.1097/md.0000000000049896
View Full Paper
Ask AI
Bookmark
Share