Skip to content

Retrospective Validation Study of a Full-Process Large Language Model Intelligent Agent Based on Multi-Center Orthopedic Electronic Health Records

Retrospective Validation Study of a Full-Process Large Language Model Intelligent Agent Based on Multi-Center Orthopedic Electronic Health Records

Status
Active, not recruiting
Phases
Unknown
Study type
Observational
Source
ChiCTR
Registry ID
ChiCTR2500115771
Enrollment
Unknown
Registered
2025-12-31
Start date
2026-01-01
Completion date
Unknown
Last updated
2026-01-05

For informational purposes only — not medical advice. Sourced from public registries and may not reflect the latest updates. Terms

Conditions

Various types of fractures

Interventions

Human-AI Comparison Groups:NA

Sponsors

Ruijin Hospital, Shanghai Jiao Tong University School of Medicine
Lead Sponsor

Eligibility

Sex/Gender
All
Age
18 Years to 120 Years

Inclusion criteria

Inclusion criteria: 1. Age >=18 years. 2. Admitted to the Orthopedics Department at Ruijin Hospital (and/or collaborating centers) for >=24 hours. 3. Possess complete records including an Admission Note, at least one Operative Record or Perioperative Management Record, and a Discharge Summary. 4. Electronic Health Record (EHR) text documents are in Chinese or a mixture of Chinese and English, suitable for text processing. 5. The EHR records contain key information related to the 11 core tasks (such as diagnosis, surgical plan, medical orders, rehabilitation suggestions, consultation opinions, etc.).

Exclusion criteria

Exclusion criteria: 1. Non-orthopedic hospitalization (e.g., patient received orthopedic consultation but was hospitalized in another department). 2. Non-formal hospitalizations, such as emergency observation or day surgery. 3. Severe absence or unclear status of critical documentation (making it impossible to determine the primary diagnosis or type of surgery). 4. Cases involving special privacy issues or legal disputes deemed unsuitable for analysis following ethical review. 5. Text that cannot be parsed due to technical reasons.

Design outcomes

Primary

MeasureTime frame
Task-level Diagnostic and Decision Accuracy;Task Correctness/Agreement Rate;

Secondary

MeasureTime frame
Model Generalization and Stability;

Countries

China

Contacts

Public ContactLei Wang

Ruijin Hospital, Shanghai Jiao Tong University School of Medicine

ray_wangs@hotmail.com+86 21 64370045

Outcome results

None listed

Source: ChiCTR (via WHO ICTRP) · Data processed: Feb 4, 2026