MétaCan
Menu
Back to cohort
Record W4413116746 · doi:10.2196/65365

Assessing the Role of Large Language Models Between ChatGPT and DeepSeek in Asthma Education for Bilingual Individuals: Comparative Study

2025· article· en· W4413116746 on OpenAlexvenueno aff
Fangfei Yu, X. Zhang, Xiaohan Tong, Kui Li, Weikuan Gu, Baiquan Yu

Bibliographic record

VenueJMIR Medical Informatics · 2025
Typearticle
Languageen
FieldMedicine
TopicArtificial Intelligence in Healthcare and Education
Canadian institutionsnot available
FundersHealth Science Center, University of Tennessee
KeywordsAsthmaComputer scienceMedicineNatural language processingPsychology

Abstract

fetched live from OpenAlex

Background: Asthma is a chronic inflammatory airway disease requiring long-term management. Artificial intelligence (AI)-driven tools such as large language models (LLMs) hold potential for enhancing patient education, especially for multilingual populations. However, comparative assessments of LLMs in disease-specific, bilingual health communication are limited. Objective: This study aimed to evaluate and compare the performance of two advanced LLMs-ChatGPT-4o (OpenAI) and DeepSeek-v3 (DeepSeek AI)-in providing bilingual (English and Chinese) education for patients with asthma, focusing on accuracy, completeness, clinical relevance, and language adaptability. Methods: A total of 53 asthma-related questions were collected from real patient inquiries across 8 clinical domains. Each question was posed in both English and Chinese to ChatGPT-4o and DeepSeek-v3. Responses were evaluated using a 7D clinical quality framework (eg, completeness, consensus consistency, and reasoning ability) adapted from Google Health. Three respiratory clinicians performed blinded scoring evaluations. Descriptive statistics and Wilcoxon signed-rank tests were applied to compare performance across domains and against theoretical maximums. Results: Both models demonstrated high overall quality in generating bilingual educational content. DeepSeek-v3 outperformed ChatGPT-4o in completeness and currency, particularly in treatment-related knowledge and symptom interpretation. ChatGPT-4o showed advantages in clarity and accessibility. In English responses, ChatGPT achieved perfect scores across 5 domains, but scored lower in clinical features (mean 3.78, SD 0.16; P=.02), treatment (mean 3.90, SD 0.05; P=.03), and differential diagnosis (mean 3.83, SD 0.29; P=.08). Conclusions: ChatGPT-4o and DeepSeek-v3 each offer distinct strengths for bilingual asthma education. While ChatGPT is more suitable for general health education due to its expressive clarity, DeepSeek provides more up-to-date and comprehensive clinical content. Both models can serve as effective supplementary tools for patient self-management but cannot replace professional medical advice. Future AI health care systems should enhance clinical reasoning, ensure guideline currency, and integrate human oversight to optimize safety and accuracy.

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 machine prediction

Teacher imitation

Not calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.

metaresearch head score (Codex)0.015
metaresearch head score (Gemma)0.053
Version: metacan-v3-hybrid-931329e0061cValidation status: machine_predicted_unvalidated
Candidate categoriesnone
Consensus categoriesnone
DomainCandidate signal: none · Consensus signal: none
Study designCandidate signal: Observational · Consensus signal: none
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.015
Threshold uncertainty score0.080

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0150.053
Meta-epidemiology (narrow)0.0010.000
Meta-epidemiology (broad)0.0010.001
Bibliometrics0.0010.001
Science and technology studies0.0010.001
Scholarly communication0.0020.002
Open science0.0010.002
Research integrity0.0010.001
Insufficient payload (model declined to judge)0.0030.001

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.

Opus teacher head0.110
GPT teacher head0.512
Teacher spread0.402 · how far apart the two teachers sit on this one work
Validation statusscore_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from it

Classification

machine, unvalidated

Machine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.

The models applied no category: nothing in the taxonomy fit this work.
Study designObservational
Domainnot available
GenreEmpirical

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".

Quick stats

Citations7
Published2025
Admission routes1
Has abstractyes

Explore more

Same venueJMIR Medical InformaticsSame topicArtificial Intelligence in Healthcare and EducationFrench-language works237,207