ChatGPT Versus DeepSeek for Breast Cancer Information Retrieval: Quantitative Comparative Study
Bibliographic record
Abstract
Background: Artificial intelligence (AI) is increasingly used to generate medical content, yet its performance in delivering clinically relevant and reliable information remains underexplored, especially in complex areas such as breast cancer. Objective: This study aimed to compare ChatGPT-4.0 and DeepSeek-V3 in generating breast cancer information, focusing on readability, content quality, and citation reliability. Methods: On the basis of publicly available patient education materials, 10 frequently asked questions were selected. Each model generated 60 responses. Three expert reviewers rated each response using a 7-point Likert scale across 5 dimensions (ie, accuracy, completeness, clarity, depth and insight, and alignment with expert answers). Readability was assessed using Flesch-Kincaid Grade Level scores. Information reliability was evaluated through interrater agreement metrics, including Cohen κ and Fleiss κ. Paired t tests were used for statistical comparisons. Results: AI models produced significantly more readable content than expert references (mean Flesch-Kincaid Grade Level difference -2.60; P<.001). ChatGPT-4.0 responses were more stylistically consistent with a median Flesch-Kincaid Grade Level score of 10.66 (IQR 0.98), whereas DeepSeek-V3 showed greater variability with a median Flesch-Kincaid Grade Level score of 10.17 (IQR 1.41). Content quality scores were DeepSeek-V3 achieving a higher mean score than ChatGPT-4.0 (6.22 [SD 0.43] vs 6.01 [SD 0.49]). In the multiresponse analysis, DeepSeek-V3 demonstrated a statistically significant advantage in accuracy (P=.041), while differences across other criteria were not statistically significant (P>.05). Human raters showed almost perfect agreement when judging source reliability (Fleiss κ=0.842 for ChatGPT's citations and 0.935 for DeepSeek's citations). Agreement between each model's citation reliability scores and the expert majority was substantial for ChatGPT (Cohen κ=0.665) and higher for DeepSeek (Cohen κ=0.782). Conclusions: Both models generated readable and clinically relevant content with comparable overall performance. ChatGPT provided more consistent readability, while DeepSeek offered more diverse references with stronger alignment to expert ratings. Continued evaluation and quality assurance are essential for the responsible clinical use of AI-generated content.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.000 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.000 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".