MétaCan
Menu
← Back to cohort
Record W4410429201 · doi:10.1101/2025.05.15.25326843

Large-scale Local Deployment of DeepSeek-R1 in Pilot Hospitals in China: A Nationwide Cross-sectional Survey

2025· preprint· en· W4410429201 on OpenAlexaff
Yuan Meng, Minghui Xu, Danli Shi, Yujian He, Yudong Xu, Weizong Wang, Weiqing Xiong, Yuting Zhao, Liuying Wang, F. X. Gan, Xiaoyu Liu, Mingguang He, Yue Qiu

Bibliographic record

VenuemedRxiv · 2025
Typepreprint
Languageen
FieldMedicine
TopicArtificial Intelligence in Healthcare and Education
Canadian institutionsArtificial Intelligence in Medicine (Canada)
FundersNational Office for Philosophy and Social Sciences
KeywordsCross-sectional studySoftware deploymentScale (ratio)ChinaGeographyBusinessRegional scienceMedicineComputer scienceCartographyArchaeology

Abstract

fetched live from OpenAlex

Summary Background The open-source release of DeepSeek-R1, a high-performing large language model (LLM), enables local deployment in Chinese hospitals. However, empirical data on deployment scale, hospital characteristics, and functional applications are lacking. Methods We conducted a nationwide cross-sectional survey of 261 hospitals in mainland China that reported local deployment of DeepSeek-R1 between Jan 1 and Mar 8, 2025. Data were collected via web-scraping from verified hospital sources and structured using a hybrid LLM-extraction pipeline. Deployment characteristics, hospital levels, regions, and model parameter distributions were analyzed using descriptive and stratified statistics. Findings DeepSeek-R1 was locally deployed in hospitals across 93·5% of Chinese provinces, with tertiary hospitals accounting for 84% of deployments. Geographical disparities were evident, with Central South, East, and North China showing higher adoption. Functional applications spanned clinical diagnosis, patient services, hospital management, and traditional Chinese medicine integration. Among hospitals disclosing model parameters, the 671B version was most prevalent (45·2%), particularly in Guangdong. Smaller models (32B, 70B) were applied in diagnosis support and intelligent Q&A, while the 671B supported more complex scenarios like strategic decision-making and quantum security. The overall deployment rate remains low nationwide (0·7%). Interpretation Local deployment of DeepSeek-R1 in China has expanded rapidly, led by high-level hospitals in economically developed regions. Model selection reflects functional demand and infrastructure capacity. DeepSeek’s broad applicability and open-source nature position it as a scalable solution for advancing AI-driven hospital transformation. However, uneven regional adoption and limited deployment in primary care suggest policy and infrastructural gaps requiring further attention. Funding This study was supported by the National Social Science Fund of China (23BGL249).

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)0.004
metaresearch head score (Gemma)0.008
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: Observational
GenreCandidate signal: Empirical · Consensus signal: Empirical
Teacher disagreement score0.038
Threshold uncertainty score0.076

Distilled classifier scores by category (both heads)

CategoryCodexGemma
Metaresearch0.0040.008
Meta-epidemiology (narrow)0.0000.000
Meta-epidemiology (broad)0.0000.000
Bibliometrics0.0010.002
Science and technology studies0.0000.001
Scholarly communication0.0010.001
Open science0.0010.001
Research integrity0.0000.001
Insufficient payload (model declined to judge)0.0030.000

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.

Opus teacher head0.096
GPT teacher head0.420
Teacher spread0.324 · 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

Citations6
Published2025
Admission routes1
Has abstractyes

Explore more

Same venuemedRxiv→Same topicArtificial Intelligence in Healthcare and Education→French-language works237,207→