大语言模型在麻醉交接班文书生成中的可行性评估

注册号:

Registration number:

ChiCTR2600127227 

最近更新日期:

Date of Last Refreshed on:

2026-06-26 16:53:01 

注册时间:

Date of Registration:

2026-06-26 00:00:00 

注册号状态:

预注册

Registration Status:

Prospective registration

注册题目:

大语言模型在麻醉交接班文书生成中的可行性评估

Public title:

Feasibility Assessment of Large Language Models in Automated Generation of Anesthesia Handover Documentation

注册题目简写:

English Acronym:

研究课题的正式科学名称:

大语言模型在麻醉交接班文书生成中的可行性评估

Scientific title:

Feasibility Assessment of Large Language Models in Automated Generation of Anesthesia Handover Documentation

研究课题代号(代码):

Study subject ID:

在二级注册机构或其它机构的注册号:

The registration number of the Partner Registry or other register:

申请注册联系人:

彭玺霖 

研究负责人:

朱涛 

Applicant:

Peng Xilin  

Study leader:

Zhu Tao 

申请注册联系人电话:

Applicant telephone:

+86 15723381536

研究负责人电话:

Study leader's
telephone:

+86 28 85423593

申请注册联系人传真 :

Applicant Fax:

研究负责人传真:

Study leader's fax:

申请注册联系人电子邮件:

Applicant E-mail:

anesopeng@163.com

研究负责人电子邮件:

Study leader's E-mail:

739501155@qq.com

申请单位网址(自愿提供):

Applicant website(voluntary supply):

研究负责人网址(自愿提供):

Study leader's website(voluntary supply):

申请注册联系人通讯地址:

中国四川省成都市武侯区国学巷37号

研究负责人通讯地址:

中国四川省成都市武侯区国学巷37号

Applicant address:

37 Guoxue Lane, Wuhou District, Chengdu, Sichuan, China

Study leader's address:

37 Guoxue Lane, Wuhou District, Chengdu, Sichuan, China

申请注册联系人邮政编码:

Applicant postcode:

研究负责人邮政编码:

Study leader's postcode:

申请人所在单位:

四川大学华西医院

Applicant's institution:

West China Hospital, Sichuan University

研究负责人所在单位:

四川大学华西医院

Affiliation of the Leader:

West China Hospital of Sichuan University

是否获伦理委员会批准:

Approved by ethic committee:

Yes

伦理委员会批件文号:

Approved No. of ethic committee:

2026年审(1372)号

伦理委员会批件附件:

Approved file of Ethical Committee:

查看附件View

批准本研究的伦理委员会名称:

四川大学华西医院生物医学伦理审查委员会

Name of the ethic committee:

Ethics Committee on Biomedical Research West China Hospital of Sichuan University

伦理委员会批准日期:

Date of approved by ethic committee:

2026-06-09 00:00:00

伦理委员会联系人:

李娜

Contact Name of the ethic committee:

Li Na

伦理委员会联系地址:

中国四川省成都市武侯区国学巷37号

Contact Address of the ethic committee:

37 Guoxue Lane, Wuhou District, Chengdu, Sichuan, China

伦理委员会联系人电话:

Contact phone of the ethic committee:

+86 28 85422654

伦理委员会联系人邮箱:

Contact email of the ethic committee:

188974152@qq.com

研究实施负责(组长)单位:

四川大学华西医院

Primary sponsor:

West China Hospital of Sichuan University

研究实施负责(组长)单位地址:

中国四川省成都市武侯区国学巷37号

Primary sponsor's address:

37 Guoxue Lane, Wuhou District, Chengdu, Sichuan, China

试验主办单位(项目批准或申办者):

Secondary sponsor:

国家:

中国

省(直辖市):

四川

市(区县):

Country:

China

Province:

Sichuan

City:

单位(医院):

四川大学华西医院

具体地址:

中国四川省成都市武侯区国学巷37号

Institution
hospital:

West China Hospital of Sichuan University

Address:

37 Guoxue Lane, Wuhou District, Chengdu, Sichuan, China

经费或物资来源:

自选课题(自筹)

Source(s) of funding:

Self-selected topic (self-funded)

研究疾病:

不适用  

Target disease:

Not Applicable

研究疾病代码:

Target disease code:

研究类型:

观察性研究

Study type:

Observational study

研究所处阶段:

其它 

Study phase:

N/A

研究设计:

队列研究 

Study design:

Cohort study 

研究目的:

本研究拟采用按亚专业分层随机抽样的设计,覆盖心脏外科、胸外科、神经外科、普通外科、骨科、泌尿外科、耳鼻喉科六个亚专业,系统评估多种主流大语言模型在麻醉交接班文书生成任务中的可行性,包括结构化信息的准确映射能力、非结构化叙事文本的表达能力以及生成效率,为麻醉信息管理的智能化提供循证依据,并为后续大规模临床验证研究奠定基础。  

Objectives of Study:

Employing a stratified random sampling strategy across six surgical subspecialties (cardiac, thoracic, neurosurgery, general, orthopedic, and otolaryngologic surgery), this study will systematically assess the feasibility of leading large language models (LLMs) in automating anesthesia handover note generation. The evaluation will focus on three core competencies: accurate mapping of structured clinical data, quality and fluency of unstructured narrative text synthesis, and operational efficiency. This comprehensive feasibility assessment will provide evidence for intelligent transformation of anesthesia information systems and establish a foundation for future multi-center clinical implementation trials.

药物成份或治疗方案详述:

 

Description for medicine or protocol of treatment in detail:

 

纳入标准:

Inclusion criteria

排除标准:

1. 急诊或限期手术; 2. 年龄<18 岁; 3. 局部麻醉或监测下麻醉(MAC)下完成的手术; 4. 手术时长<1 小时; 5. 术后未经 PACU/ICU 直接返回病房(无正式麻醉交接场景); 6. 病历关键模块缺失(任意一项缺失即排除); 7. 同一患者多次手术,仅纳入首次手术病历。

Exclusion criteria:

1. Emergency or time-limited surgery; 2. Age <18 years old; 3. Surgeries performed under local anesthesia or monitored anesthesia (MAC); 4. The operation duration is less than 1 hour; 5. Return to the ward directly without a PACU/ICU after the operation (without a formal anesthesia handover scene); 6. Missing key modules of the medical record (any missing item is excluded); 7. For the same patient who has undergone multiple surgeries, only the medical record of the first surgery will be included.

研究实施时间:

Study execute time:

From 2026-06-12 00:00:00 To 2027-06-12 00:00:00  

征募观察对象时间:

Recruiting time:

From 2026-07-01 00:00:00 To 2027-06-12 00:00:00

干预措施:

Interventions:

组别:

观察组

样本量:

70

Group:

Observation group

Sample size:

干预措施:

干预措施代码:

Intervention:

none

Intervention code:

研究实施地点:

Countries of recruitment and research settings:

国家:

中国

省(直辖市):

四川 

市(区县):

 

Country:

China

Province:

Sichuan

City:

单位(医院):

四川大学华西医院 

单位级别:

三甲 

Institution
hospital:

West China Hospital of Sichuan University

Level of the institution:

Tertiary A

测量指标:

Outcomes:

指标中文名:

大语言模型生成医疗文书的结构保真度

指标类型:

主要指标

Outcome:

The structural fidelity of medical documents generated by large language models

Type:

Primary indicator

测量时间点:

所有医疗文书生成后

测量方法:

结构覆盖率(Coverage): 又称精确交并比,评估成功匹配的结构占模型生成的所有结构与标准要求的总和(并集)的比例② 召回率 (Recall): 评估标准要求中有多少被成功生成。③ 含义一致性 (Consistency): 又称软交并比,反映大体结构的语义贴合程度。④ 模糊召回率 (Recallfuzzy):又称软召回率,评估允许表达差异下,预期结构被成功包含的比例。

Measure time point of outcome:

End of all medical documents are generated

Measure method:

Coverage: Also known as the precise intersection-over-union, it assesses the proportion of successfully matched structures against the total sum (union) of all structures generated by the model and the standard requirements.② Recall: Evaluates how many of the standard requirements were successfully generated.③ Consistency: Also known as soft intersection-over-union, it reflects the degree of semantic alignment of the overall structure.④Recallfuzzy: Also known as soft recall, it assesses the pr

指标中文名:

大语言模型生成医疗文书的内容准确性

指标类型:

主要指标

Outcome:

The content accuracy of medical documents generated by large language models

Type:

Primary indicator

测量时间点:

所有医疗文书生成后

测量方法:

应用率 (Application Rate):评估大模型按照规定模板输出字段的执行力。② 正确率 (Accuracy Rate):在模型提及的所有字段中,内容与真实病历一致的比例。③ 幻觉率 (Hallucination Rate):在模型提及的所有字段中,发生事实错误或幻觉的比例。

Measure time point of outcome:

End of all medical documents are generated

Measure method:

Application Rate: Assesses the execution of the large model in outputting fields according to the specified template. ② **Accuracy Rate**: The proportion of content among all fields mentioned by the model that is consistent with the actual medical records. ③ **Hallucination Rate**: The proportion of factual errors or hallucinations occurring among all fields mentioned by the model.

指标中文名:

大语言模型非结构化文本的表达能力

指标类型:

主要指标

Outcome:

The expressive capability of unstructured text generated by large language models.

Type:

Primary indicator

测量时间点:

所有医疗文书生成后

测量方法:

Completeness(完整性):交接关键要点是否齐全② Correctness(正确性):是否与病历一致(尤其数值/药物/事件)③ Conciseness(简洁性):是否冗余,能否快速抓重点④ Clinical utility(临床可用性):接班麻醉医师是否“拿来能用”⑤ Alertness(安全警示性):高风险信息是否被主动突出⑥ Actionability(可执行性):是否给

Measure time point of outcome:

End of all medical documents are generated

Measure method:

Completeness: Whether the key points of the handover are comprehensive. ② **Correctness**: Whether the information is consistent with the medical records (especially numerical values, medications, and events). ③ **Conciseness**: Whether the text is redundant and if the key points can be quickly grasped. ④ **Clinical Utility**: Whether the incoming anesthesiologist can "use it directly." ⑤ **Alertness**: Whether high-risk information is actively highlighted. ⑥ **Actionability**: W

指标中文名:

大语言模型输出时间

指标类型:

次要指标

Outcome:

Output time of large language models.

Type:

Secondary indicator

测量时间点:

生成文书时

测量方法:

计秒器记录不同模型输出的时间(T1:prompt1列名筛选耗时,T2:prompt2 报告生成耗时,Ttotal:T1+T2)

Measure time point of outcome:

Output time during document generation.

Measure method:

A stopwatch records the output times of different models: T1: Time taken for prompt 1 column name filtering. T2: Time taken for report generation with prompt 2. Ttotal: Total time, calculated as T1 + T2.

采集人体标本:

Collecting sample(s)
from participants:

标本中文名:

组织:

Sample Name:

NA

Tissue:

人体标本去向

其它  

说明

Fate of sample:

0thers  

Note:

征募研究对象情况:

Recruiting status:

尚未开始

Not yet recruiting

年龄范围:

Participant age:

最小 Min age 18 years
最大 Max age years

性别:

男女均可

Gender:

Both

随机方法(请说明由何人用什么方法产生随机序列):

Randomization Procedure (please state who generates the random number sequence and by what method):

None

是否公开试验完成后的统计结果:

Calculated Results after the Study Completed public access:

不公开/Private

盲法:

Blinding:

None

是否共享原始数据:

IPD sharing

否No

共享原始数据的方式(说明:请填入公开原始数据日期和方式,如采用网络平台,需填该网络平台名称和网址):

不共享

The way of sharing IPD”(include metadata and protocol, If use web-based public database, please provide the url):

No share

数据采集和管理(说明:数据采集和管理由两部分组成,一为病例记录表(Case Record Form, CRF),二为电子采集和管理系统(Electronic Data Capture, EDC),如ResMan即为一种基于互联网的EDC:

病例报告表,电子采集和管理系统

Data collection and Management (A standard data collection and management system include a CRF and an electronic data capture:

Case Report Form and Electronic Data Capture

数据与安全监察委员会:

Data and Safety Monitoring Committee:

有/Yes

注册人:

Name of Registration:

 2026-06-26 16:53:01