Table 1_AI dialogues in cartilage repair: which guides evidence-based decisions better?.do

A study comparing the performance of ChatGPT (GPT-4), DeepSeek (V3), and Google Search on cartilage repair questions. The dataset includes 20 frequently asked questions from cartilage tissue engineering (2023) and cartilage repair surgery (2024) domains, with answers evaluated for accuracy, safety, and readability. Authored by Sen Yang Xiao and last updated on 2026-05-29.

Use Cases

• Benchmarking AI model accuracy and safety in medical domains based on the described Accuracy-Safety-Hallucination (ASH) framework.
• Analyzing the readability of AI-generated medical answers compared to search engine results based on Flesch-Kincaid Grade Level scores.
• Studying question classification patterns in medical FAQs using the modified Rothwell taxonomy mentioned in the methods.

Strengths

• Evaluation based on a dual-axis framework integrating classification, blinded quality scoring, and readability analysis.
• Results include specific median accuracy scores (e.g., 5.00 for DeepSeek in CTE) and statistical significance (p-values) for comparisons.
• Independent evaluation by three blinded raters for answer quality assessment.

Limitations

• Dataset scope is limited to 20 questions across two specific medical subdomains.
• The primary data file is a 20.1 KB DOCX document; the underlying structured evaluation data is not directly available.
• Column-level documentation is absent; field semantics must be inferred from the study description.

Provenance

Source

figshare, authored by Sen Yang Xiao.

Collection Method

Questions were sourced from Google Search top FAQs; answers were generated by ChatGPT, DeepSeek, and Google Search for a matched three-way comparison.

Time Range

Questions sourced from cartilage tissue engineering (2023) and cartilage repair surgery (2024) domains.

Freshness

Last updated 2026-05-29 05:53:28; freshness should be verified.