ChiCTR2600122315 版本V1.0 版本创建时间2026/04/12 22:03:56 中国临床试验注册中心

审核状态:

Project audit state:

通过审核

Successful

注册号:

Registration number:

ChiCTR2600122315 

最近更新日期:

Date of Last Refreshed on:

2026-04-12 22:03:51 

注册时间:

Date of Registration:

2026-04-12 00:00:00 

注册号状态:

补注册

Registration Status:

Retrospective registration

注册题目:

DeepSeek 模型辅助胃癌临床决策的回顾性评估

Public title:

Retrospective Evaluation of DeepSeek Model in Assisting Clinical Decision-Making for Gastric Cancer

注册题目简写:

English Acronym:

研究课题的正式科学名称:

DeepSeek 大模型辅助胃癌临床决策的回顾性效能评估研究

Scientific title:

Retrospective Evaluation of the Efficacy of DeepSeek Large Language Model (LLM) in Assisting Clinical Decision-Making for Gastric Cancer

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:

陈韬 

研究负责人:

陈韬 

Applicant:

Tao chen 

Study leader:

Tao Chen 

申请注册联系人电话:

Applicant telephone:

+86 797 5889720

研究负责人电话:

Study leader's telephone:

+86 797 5889157

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:

linchuangyanjiu999@163.com

研究负责人电子邮件:

Study leader's E-mail:

13911111111@163.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:

中国江西省赣州市章贡区梅关大道16号

研究负责人通讯地址:

梅关大道16号

Applicant address:

No. 16 Meiguan Avenue, Zhanggong District, Ganzhou City, Jiangxi Province, China

Study leader's address:

16 Meiguan Avenue, Ganzhou, Jiangxi

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:

南方医院赣州医院(赣州市人民医院)

Applicant's institution:

Southern Hospital Ganzhou Hospital (Ganzhou People's Hospital)

研究负责人所在单位:

赣州市人民医院

Affiliation of the Leader:

ganzhou people's hospital

是否获伦理委员会批准:

是/Yes

Approved by ethic committee:

Yes

伦理委员会批件文号:

Approved No. of ethic committee:

PJB2025-319-01

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:

赣州市人民医院伦理委员会

Name of the ethic committee:

Ethics Committee of Ganzhou People's Hospital

伦理委员会批准日期:

Date of approved by ethic committee:

2025-09-10 00:00:00

伦理委员会联系人:

刘文群

Contact Name of the ethic committee:

Liu WenQun

伦理委员会联系地址:

梅关大道16号

Contact Address of the ethic committee:

16 Meiguan Avenue, Ganzhou, Jiangxi

伦理委员会联系人电话:

Contact phone of the ethic committee:

+86 797 5889157

伦理委员会联系人邮箱:

Contact email of the ethic committee:

408488245@qq.com

研究实施负责(组长)单位:

赣州市人民医院

Primary sponsor:

ganzhou people's hospital

研究实施负责(组长)单位地址:

梅关大道16号

Primary sponsor's address:

16 Meiguan Avenue, Ganzhou, Jiangxi

试验主办单位(项目批准或申办者):

Secondary sponsor:

国家:

中国

省(直辖市):

江西省

市(区县):

Country:

China

Province:

Jiangxi

City:

单位(医院):

赣州市人民医院

具体地址:

梅关大道16号

Institution
hospital:

ganzhou people's hospital

Address:

16 Meiguan Avenue, Ganzhou, Jiangxi

经费或物资来源:

自选课题(自筹)

Source(s) of funding:

Self-raised

Target disease:

Gastric cancer

Target disease code:

研究类型:

观察性研究

Study type:

Observational study

研究所处阶段:

回顾性研究 

Study phase:

Retrospective study

研究设计:

队列研究 

Study design:

Cohort study 

研究目的:

主要目的: 验证 DeepSeek 胃癌大模型在回顾性队列中辅助临床决策的综合效能: (1)诊断准确性(vs 病理金标准); (2)治疗规划与指南符合率(vs CSCO 2024)。 次要目的: (1)评估模型对检查项目合理化推荐的贡献(减少冗余检查率); (2)分析模型辅助决策对诊疗效率的影响(诊断至治疗决策时间缩短比例); (3)探索模型在特殊人群(老年/晚期/HER2+)中的适用性。  

Objectives of Study:

Primary Objectives To verify the comprehensive efficacy of the DeepSeek Gastric Cancer Large Model in assisting clinical decision-making within a retrospective cohort:(1) Diagnostic accuracy (compared with the pathological gold standard);(2) Compliance rate of treatment planning with clinical guidelines (compared with CSCO Guidelines 2024). Secondary Objectives (1) To evaluate the model’s contribution to the rational recommendation of examination items (by reducing the rate of redundant examinations);(2) To analyze the impact of model-assisted decision-making on diagnostic and therapeutic efficiency (reduction ratio of the time from diagnosis to treatment decision-making);(3) To explore the applicability of the model in special populations (elderly patients, advanced-stage gastric cancer patients, HER2-positive gastric cancer patients).

药物成份或治疗方案详述:

 

Description for medicine or protocol of treatment in detail:

 

纳入标准:

1.年龄>=18 岁; 2014 年 1 月-2023 年 12 月就诊于赣州市人民医院并接受治疗(手术/化疗/ 靶向/放疗); 2.病历包含完整基线资料(实验室检查结果、胃镜/手术病理报告、影像学报 告、完整病历记录、随访记录); 3.豁免知情同意:数据脱敏处理,距末次随访>1 年。

Inclusion criteria

1.Age >= 18 years; 2.Attended Ganzhou People's Hospital and received treatment (surgery, chemotherapy, targeted therapy, or radiotherapy) from January 2014 to December 2023; 3.Complete baseline data available in medical records, including laboratory test results, gastroscopy/surgical pathological reports, imaging reports, complete medical records, and follow-up records; 4.Waiver of informed consent: Data de-identification has been performed, and the time from the last follow-up is > 1 year.

排除标准:

1.病历关键数据缺失(如无病理分期、未记录治疗方案);
2.合并其他恶性肿瘤病史;
3.参与同期干预性临床试验者;
4.经研究者判断认为不适合参与本研究的患者。

Exclusion criteria:

1.Missing key data in medical records (e.g., no pathological staging, unrecorded treatment plan);
2.History of other concurrent malignant tumors;
3.Participation in concurrent interventional clinical trials;
4.Patients deemed unsuitable for participation in this study by the researchers.

研究实施时间:

Study execute time:

From 2025-09-10 00:00:00 To 2026-06-30 00:00:00  

征募观察对象时间:

Recruiting time:

From 2025-09-10 00:00:00 To 2026-06-30 00:00:00  

干预措施:

Interventions:

组别:

训练集

样本量:

1500

Group:

Training Set

Sample size:

干预措施:

干预措施代码:

Intervention:

None

Intervention code:

组别:

验证集

样本量:

500

Group:

Validation Set

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):

江西省 

市(区县):

 

Country:

China 

Province:

Jiangxi 

City:

 

单位(医院):

赣州市人民医院 

单位级别:

三级甲等 

Institution
hospital:

ganzhou people's hospital

Level of the institution:

Tertiary A

测量指标:

Outcomes:

指标中文名:

检查推荐合理性、诊疗效率、安全性

指标类型:

次要指标

Outcome:

rationality of examination recommendations, diagnostic and therapeutic efficiency, and safety

Type:

Secondary indicator

测量时间点:

所有病历数据提取截止至 2024 年 12 月 31 日

测量方法:

1. 提取 “模型检查推荐”:获取模型对每例患者的完整推荐信息,包括推荐检查项目、推荐理由、检查优先级; 2. 制定 “合理性判定标准”:制定三级判定标准,包括完全合理、部分合理、不合理; 3. 计算 “检查推荐合理率”:完全合理率 =(完全合理例数)/(总可评估例数)×100%,总合理率(完全 + 部分)=(完全合理例数 + 部分合理例数)/(总可评估例数)×100%; 4. 收集 “时间数据”

Measure time point of outcome:

All medical record data extraction is cut off on December 31, 2024.

Measure method:

Extract the model’s exam recommendations: complete info per patient (items, reasons, priority);Formulate 3-level rationality criteria (complete/partial rationality, irrationality);Calculate rates: complete rate=(complete cases/total evaluable)×100%; total rate=(complete+partial cases/total evaluable)×100%; Collect time data: timestamps from EMR and exam appointment systems; Analyze efficiency differences: compare model-assisted vs baseline control groups’ time indicators (median/IQR for skewed d

指标中文名:

模型诊断与金标准一致性、治疗推荐指南依从率

指标类型:

主要指标

Outcome:

Consistency between the model's diagnosis and the gold standard, compliance rate of treatment recommendations with clinical guidelines

Type:

Primary indicator

测量时间点:

所有病历数据提取截止至 2024 年 12 月 31 日

测量方法:

1.数据收集:模型诊断结果、金标准结果; 2. 结果整理:将模型结果与金标准结果整理为四格表,包括真阳性(TP)、假阳性(FP)、真阴性(TN)、假阴性(FN); 3.统计一致性指标:计算准确性、灵敏度、特异度、阳性预测值、阴性预测值、Kappa 系数、ROC 曲线与 AUC; 4.全面评估一致性:提取 “模型治疗推荐”、确定 “指南推荐标准”、判定 “依从性等级”、计算 “指南依从率”。

Measure time point of outcome:

All medical record data extraction is cut off on December 31, 2024.

Measure method:

Data Collection: Model's diagnostic results and gold standard results;Result Collation: Collate the model's results and gold standard results into a 2×2 contingency table;Statistical Consistency Indicators: Calculate accuracy, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), Kappa coefficient, ROC curve, and AUC;Comprehensive Assessment of Consistency: Extract the "model's treatment recommendations", define the "guideline-recommended standards", determi

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:

正在进行

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

Yes

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):

国家生物信息中心 https://ngdc.cncb.ac.cn/gsub/

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

China National center for Bioinformation (https://ngdc.cncb.ac.cn/gsub/)

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:

病例报告表上有关受试者数据应以受试者编码方式记录,受试者只能通过受试者编码或其姓名首字母缩写识别。 本研究采用数据库的形式进行数据管理,从数据录入到源数据的核查要求到质控数据的质疑解答,最后到数据锁定及导出的操作、确认数据无疑问后,各方签署数据库锁定申请表,由数据管理员对数据库进行锁定。数据库锁定后,由数据管理员导出分析数据库,交统计人员进行统计分析。锁定后的数据不可再编辑,数据库锁定之后发现的问题,经确认后可在统计分析程序中修正。

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

Data related to participants on the Case Report Form (CRF) shall be recorded using participant codes. Participants can only be identified by their participant codes or the initials of their name. For this study, a database is adopted for data management. The process includes data entry, source data verification, resolution of queries related to quality control data, and finally the operations of data locking and export. After confirming that there are no queries regarding the data, all relevant parties shall sign the Database Lock Application Form, and the data manager shall lock the database. After the database is locked, the data manager will export the analysis database and submit it to statisticians for statistical analysis. Locked data cannot be edited further; any issues identified after database locking may be corrected in the statistical analysis program upon confirmation.

数据与安全监察委员会:

Data and Safety Monitoring Committee:

无/No

注册人:

Name of Registration:

 2026-04-12 22:03:51