DeepSeek-R1 辅助临床医师开展对脑胶质瘤鉴别诊断效果的随机对照试验研究

注册号:

Registration number:

ChiCTR2600124585 

最近更新日期:

Date of Last Refreshed on:

2026-05-14 10:20:49 

注册时间:

Date of Registration:

2026-05-14 00:00:00 

注册号状态:

预注册

Registration Status:

Prospective registration

注册题目:

DeepSeek-R1 辅助临床医师开展对脑胶质瘤鉴别诊断效果的随机对照试验研究

Public title:

A Randomized Controlled Trial of DeepSeek-R1 Assistance for Clinicians in the Differential Diagnosis of Glioma

注册题目简写:

English Acronym:

研究课题的正式科学名称:

DeepSeek-R1 辅助临床医师开展对脑胶质瘤鉴别诊断效果的随机对照试验研究

Scientific title:

A Randomized Controlled Trial of DeepSeek-R1 Assistance for Clinicians in the Differential Diagnosis of Glioma

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:

李军旗 

研究负责人:

李军旗 

Applicant:

Li Junqi 

Study leader:

Li Junqi 

申请注册联系人电话:

Applicant telephone:

+86 18823215364

研究负责人电话:

Study leader's
telephone:

+86 755 83923333

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:

770948499@qq.com

研究负责人电子邮件:

Study leader's E-mail:

770948499@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:

中国广东省深圳市福田区莲花路1120号

研究负责人通讯地址:

中国广东省深圳市福田区莲花路1120号

Applicant address:

No. 1120, Lianhua Road, Futian District, Shenzhen, Guangdong, China

Study leader's address:

No. 1120, Lianhua Road, Futian District, Shenzhen, Guangdong, 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:

北京大学深圳医院

Applicant's institution:

Peking University Shenzhen Hospital

研究负责人所在单位:

北京大学深圳医院

Affiliation of the Leader:

Peking University Shenzhen Hospital

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:

北大深医伦审(研)[2025]第(302)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:

北京大学深圳医院科研伦理审查委员会

Name of the ethic committee:

Research Ethics Review Committee of Peking University Shenzhen Hospital

伦理委员会批准日期:

Date of approved by ethic committee:

2025-12-09 00:00:00

伦理委员会联系人:

杨珍妮

Contact Name of the ethic committee:

Jenny Yang

伦理委员会联系地址:

中国广东省深圳市福田区莲花路1120号

Contact Address of the ethic committee:

No. 1120, Lianhua Road, Futian District, Shenzhen, Guangdong, China

伦理委员会联系人电话:

Contact phone of the ethic committee:

+86 755 83923333

伦理委员会联系人邮箱:

Contact email of the ethic committee:

3125392358@qq.com

研究实施负责(组长)单位:

北京大学深圳医院

Primary sponsor:

Peking University Shenzhen Hospital

研究实施负责(组长)单位地址:

中国广东省深圳市福田区莲花路1120号

Primary sponsor's address:

No. 1120, Lianhua Road, Futian District, Shenzhen, Guangdong, 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):

广东

市(区县):

Country:

China

Province:

Guangdong

City:

单位(医院):

北京大学深圳医院

具体地址:

中国广东省深圳市福田区莲花路1120号

Institution
hospital:

Peking University Shenzhen Hospital

Address:

No. 1120, Lianhua Road, Futian District, Shenzhen, Guangdong, 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:

无  

Target disease:

None

研究疾病代码:

Target disease code:

研究类型:

干预性研究

Study type:

Interventional study

研究所处阶段:

其它 

Study phase:

N/A

研究设计:

随机平行对照 

Study design:

Parallel 

研究目的:

本研究为随机对照试验,旨在评估人工智能辅助这一新型干预措施对临床医师鉴别诊断正确率的影响。研究对象为神经外科医师参与者,研究者将其随机分配至AI辅助组和非AI辅助组。AI辅助组在常规诊断资源基础上,额外获得由研究者依据标准化提示词工程输入DeepSeek-R1后生成的推理过程、相关疾病诊断结果及诊断依据等辅助信息;非AI辅助组仅使用常规诊断资源完成判断。研究主要比较两组医师在标准化病例任务中的Top 1诊断正确率差异,并进一步评价诊断时间、诊断质量评分、LLM幻觉发生率及使用体验。病例资料仅作为统一的标准化诊断任务材料,不作为本研究的受试者。本研究重点在于评价AI辅助干预对医师诊断决策效果的影响,而非评价某一诊断工具本身的敏感度、特异度或诊断效能。  

Objectives of Study:

This study is a randomized controlled trial designed to evaluate the effect of an artificial intelligence assistance intervention on clinicians’ diagnostic accuracy in differential diagnosis tasks. The study participants are neurosurgical physicians, who will be randomly assigned to an AI-assisted group or a non-AI-assisted group. In the AI-assisted group, physicians will receive conventional diagnostic resources plus supplementary information generated by DeepSeek-R1 based on standardized prompt engineering prepared by the investigators, including reasoning processes, candidate diagnoses, and supporting diagnostic rationales. Physicians in the non-AI-assisted group will complete the same tasks using conventional diagnostic resources only. The primary objective is to compare the Top 1 diagnostic accuracy between the two groups in standardized case-based tasks, with additional evaluation of diagnostic time, diagnostic quality score, LLM hallucination rate, and user experience. Case materials serve only as standardized task materials and are not considered study participants. The focus of this study is to assess the effect of an AI-assisted intervention on physicians’ diagnostic decision-making, rather than to evaluate the sensitivity, specificity, or intrinsic diagnostic performance of a diagnostic tool itself.

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:

本研究病例库中病例的原主管医师或对病例信息有预先了解者。

Exclusion criteria:

The former responsible physicians or those with prior knowledge of the case information in the medical database of this study.

研究实施时间:

Study execute time:

From 2025-12-01 00:00:00 To 2026-11-30 00:00:00  

征募观察对象时间:

Recruiting time:

From 2026-05-30 00:00:00 To 2026-06-30 00:00:00

干预措施:

Interventions:

组别:

AI辅助组

样本量:

10

Group:

AI-Assisted Group

Sample size:

干预措施:

AI辅助

干预措施代码:

Intervention:

AI-Assisted

Intervention code:

组别:

非 AI辅助组

样本量:

10

Group:

Non-AI Assisted 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):

广东 

市(区县):

 

Country:

China

Province:

Guangdong

City:

单位(医院):

北京大学深圳医院 

单位级别:

三级甲等 

Institution
hospital:

Peking University Shenzhen Hospital

Level of the institution:

Tertiary A

测量指标:

Outcomes:

指标中文名:

Top 1 诊断准确率

指标类型:

主要指标

Outcome:

Top 1 diagnostic accuracy rate

Type:

Primary indicator

测量时间点:

测量方法:

医师或模型给出的最可能诊断(Top 1诊断)与金标准(病理学)一致的病例比例。

Measure time point of outcome:

Measure method:

The proportion of cases in which the most likely diagnosis (Top 1 diagnosis) given by the physician or model matches the gold standard (pathology).

指标中文名:

诊断时间

指标类型:

次要指标

Outcome:

Diagnostic Time

Type:

Secondary indicator

测量时间点:

从医师接收到完整病例信息到做出最终Top 1诊断所花费的时间(以秒 计)。

测量方法:

从医师接收到完整病例信息到做出最终Top 1诊断所花费的时间(以秒计)。

Measure time point of outcome:

The time of the physician receiving complete case information to making the final Top 1 .diagnosis.

Measure method:

The time elapsed (in seconds) from the physician receiving complete case information to making the final Top 1 diagnosis.

指标中文名:

LLM幻觉发生率

指标类型:

次要指标

Outcome:

LLM Hallucination Rate

Type:

Secondary indicator

测量时间点:

最终诊断结果

测量方法:

DeepSeek-R1在独立诊断过程中,出现“提供与病例事实明显不符 的信息”或“在推理过程中引入未经证实的医学概念”的比例。

Measure time point of outcome:

Final diagnosis

Measure method:

Defined as the proportion of DeepSeek-R1 outputs containing factual inaccuracies relative to case information or introducing unverified medical concepts.

指标中文名:

诊断质量评分

指标类型:

次要指标

Outcome:

Diagnostic Quality Score

Type:

Secondary indicator

测量时间点:

最终诊断结

测量方法:

采用5分制Ordinal量表对生成的鉴别诊断列表进行评分(5分:包含确切诊断;4分:包含非常接近的诊断;3分:包含有提示意义的诊断;2分:包含相关但无帮助的诊断;0分:完全不相关)

Measure time point of outcome:

Final diagnosis

Measure method:

A 5-point ordinal scale is used to score the generated differential diagnosis list (5 points: includes the definitive diagnosis; 4 points: includes a very close diagnosis; 3 points: includes a suggestive diagnosis; 2 points: includes related but unhelpful diagnoses; 0 points: completely irrelevant)

指标中文名:

诊断效能指标

指标类型:

次要指标

Outcome:

Diagnostic Performance Metrics

Type:

Secondary indicator

测量时间点:

从医师接收到完整病例信息到做出最终Top 1诊断所花费的时间(以秒 计)。

测量方法:

采用5分制Ordinal量表对生成的鉴别诊断列表进行评分(5分:包含确切诊断;4分:包含非常接近的诊断;3分:包含有提示意义的诊断;2分:包含相关但无帮助的诊断;0分:完全不相关)

Measure time point of outcome:

Diagnostic Time

Measure method:

A 5-point ordinal scale is used to score the generated differential diagnosis list (5 points: includes the definitive diagnosis; 4 points: includes a very close diagnosis; 3 points: includes a suggestive diagnosis; 2 points: includes related but unhelpful diagnoses; 0 points: completely irrelevant).

采集人体标本:

Collecting sample(s)
from participants:

标本中文名:

组织:

Sample Name:

N/A

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 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):

随机序列由不参与研究对象招募、分组实施及结果评估的独立统计学人员采用区组随机化方法生成。

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

The random sequence was generated by an independent statistician who was not involved in participant recruitment, intervention implementation, or outcome assessment, using block randomization.

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:

单盲,对评估者隐藏分组

Blinding:

Single blind study with blinded-evaluators

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):

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

None

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:

伦理与数据安全:研究遵循《赫尔辛基宣言》,所有数据脱敏处理,保护患者和医师隐私。

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

Ethics and Data Security: The study will adhere to the Declaration of Helsinki. All data will be de-identified to protect patient and physician privacy.

数据与安全监察委员会:

Data and Safety Monitoring Committee:

有/Yes

注册人:

Name of Registration:

 2026-05-14 10:20:34