ChiCTR2600117374 版本V1.0 版本创建时间2026/01/23 08:28:33 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：

ChiCTR2600117374 

最近更新日期：

Date of Last Refreshed on：

2026-01-23 08:28:29 

注册时间：

Date of Registration：

2026-01-23 00:00:00 

注册号状态：

预注册

Registration Status：

Prospective registration

注册题目：

基于CT深度学习算法预测FEV?与FVC的多中心随机试验：非劣效性及对临床决策影响的两阶段评估

Public title：

A multicenter randomized trial based on CT deep learning algorithm for predicting FEV? and FVC: Two-stage evaluation of non-inferiority and impact on clinical decision-making

注册题目简写：

English Acronym：

研究课题的正式科学名称：

基于CT深度学习算法预测FEV?与FVC的多中心随机试验：非劣效性及对临床决策影响的两阶段评估

Scientific title：

A multicenter randomized trial based on CT deep learning algorithm for predicting FEV? and FVC: Two-stage evaluation of non-inferiority and impact on clinical decision-making

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：

邢婉婷 

研究负责人：

边云 

Applicant：

Xing Wanting  

Study leader：

Bian Yun  

申请注册联系人电话：

Applicant telephone：

+86 185 6765 8139

研究负责人电话：

Study leader's telephone：

+86 138 1635 7024

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：

yjlssxwt@163.com

研究负责人电子邮件：

Study leader's E-mail：

bianyun2012@foxmail.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：

上海市杨浦区长海路168号

研究负责人通讯地址：

上海市杨浦区长海路168号

Applicant address：

No. 168, Changhai Road, Yangpu, ShanghaiNo. 168, Changhai Road, Yangpu, Shanghai

Study leader's address：

No. 168, Changhai Road, Yangpu, ShanghaiNo. 168, Changhai Road, Yangpu, Shanghai

申请注册联系人邮政编码：

Applicant postcode：

223001

研究负责人邮政编码：

Study leader's postcode：

223001

申请人所在单位：

海军军医大学长海医院

Applicant's institution：

Changhai Hospital, The Navy Military Medical University

研究负责人所在单位：

海军军医大学长海医院

Affiliation of the Leader：

Changhai Hospital, The Navy Military Medical University

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：

CHEC-Y2026-077

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：

上海长海医院伦理委员

Name of the ethic committee：

Shanghai Changhai Hospital Ethics Committee

伦理委员会批准日期：

Date of approved by ethic committee：

2026-01-10 00:00:00

伦理委员会联系人：

张优琴

Contact Name of the ethic committee：

Zhang Youqin

伦理委员会联系地址：

上海市杨浦区长海路168号

Contact Address of the ethic committee：

No.168, Changhai Road, Yangpu District, Shanghai

伦理委员会联系人电话：

Contact phone of the ethic committee：

+86 21 3116 2338

伦理委员会联系人邮箱：

Contact email of the ethic committee：

研究实施负责（组长）单位：

海军军医大学长海医院

Primary sponsor：

Changhai Hospital, The Navy Military Medical University

研究实施负责（组长）单位地址：

上海市杨浦区长海路168号

Primary sponsor's address：

No. 168, Changhai Road, Yangpu, Shanghai

试验主办单位(项目批准或申办者)：

Secondary sponsor：

国家：

中国

省(直辖市)：

上海

市(区县)：

Country：

China

Province：

Shanghai

City：

单位(医院)：

海军军医大学长海医院

具体地址：

上海市杨浦区长海路168号

Institution
hospital：

Changhai Hospital, The Navy Military Medical University

Address：

No. 168, Changhai Road, Yangpu, Shanghai

经费或物资来源：

自筹

Source(s) of funding：

Self-raised

Target disease：

Chronic Obstructive Pulmonary Disease

Target disease code：

研究类型：

诊断试验

Study type：

Diagnostic test

研究所处阶段：

 其它 

Study phase：

N/A

研究设计：

诊断试验诊断准确性 

Study design：

Diagnostic test for accuracy 

研究目的：

1. 主要目的 本研究的核心目的在于检验AI算法预测FEV?与FVC绝对值的表现是否非劣于金标准肺功能检查： FEV1的均值差（AI预测值 - 肺功能实测值）的95%置信区间（CI）上限须 <= 0.15升 FVC的均值差（AI预测值 - 肺功能实测值）的95%置信区间（CI）上限须 <= 0.20升 2. 次要目的 量化AI预测信息的可见性对临床医师决策行为及整体诊疗流程效率的实际影响，包括： 由AI预测值衍生的GOLD分级与肺功能检查结果的一致性（采用二次加权Kappa系数）； 在两阶段评估中临床决策的改变率与改变幅度； 诊疗流程所用时间；患者报告的就诊体验； 以及算法的一致性与校准度（通过一致性相关系数CCC、Bland-Altman分析及预测区间覆盖率进行评估）。  

Objectives of Study：

1. Main objective The core aim of this study is to verify whether the performance of AI algorithms in predicting FEV? and FVC absolute values is non-inferior to the gold standard pulmonary function test: The upper limit of the 95% confidence interval (CI) for the mean difference (AI prediction value - actual pulmonary function measurement) of FEV1 must be <= 0.15 liters The upper limit of the 95% confidence interval (CI) for the mean difference (AI prediction value - actual pulmonary function measurement) of FVC must be <= 0.20 liters 2. Secondary objectives To quantify the actual impact of the visibility of AI predictive information on the decision-making behavior of clinicians and the efficiency of the overall diagnostic process, including: The consistency between the GOLD classification derived from AI prediction values and the results of pulmonary function tests (using the secondary weighted Kappa coefficient); The rate and magnitude of changes in clinical decisions in the two-stage assessment; The time spent on the diagnostic process; The patient-reported experience of the visit; And the consistency and calibration of the algorithm (evaluated through the consistency correlation coefficient CCC, Bland-Altman analysis, and coverage of the prediction interval).

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至90岁 2. 计划接受胸部CT检查且同日或前后24小时内可完成肺功能检查 3. 具备知情同意能力 4. 经医师评估可尝试屏气训练

Inclusion criteria

1. Age range: 18 to 90 years old 2. Plan to undergo chest CT examination and can complete pulmonary function test on the same day or within 24 hours thereafter 3. Have the ability to give informed consent 4. Can be evaluated by the physician to attempt diaphragmatic breathing training

排除标准：

1. 急性呼吸衰竭：参照ATS/ERS等指南，定义为室内空气条件下动脉血氧分压（PaO2）<60 mmHg或呼吸频率>30次/分，或需有创通气支持 2. 近14天内接受过重大胸部手术 3. 妊娠或哺乳期女性 4. CT图像质量不合格（存在重度金属伪影、运动伪影或吸气明显不足） 5. 存在肺功能检查的绝对禁忌证或经医师判定不宜进行检查

Exclusion criteria：

1. Acute respiratory failure: According to the guidelines of ATS/ERS and others, it is defined as arterial partial pressure of oxygen (PaO2) < 60 mmHg in indoor air conditions or respiratory rate > 30 breaths per minute, or requiring invasive ventilation support. 2. Within the past 14 days, has undergone major thoracic surgery. 3. Pregnant or lactating women. 4. Inadequate quality of CT images (presence of severe metal artifacts, motion artifacts, or significant insufficient inhalation). 5. Has absolute contraindications for pulmonary function tests or has been judged by the physician not suitable for the test.

研究实施时间：

Study execute time：

从 From 2026-01-20 00:00:00至 To 2027-12-01 00:00:00  

征募观察对象时间：

Recruiting time：

从From 2026-01-25 00:00:00 至 To 2027-06-01 00:00:00  

诊断试验：

Diagnostic Tests：

金标准或参考标准（即可准确诊断某疾病的单项方法或多项联合方法，在本研究中用于诊断是否有该病的临床参考标准）：

金标准肺功能检查

Gold Standard or Reference Standard (The clinical reference standards required to establish the presence or absence of the target condition in the tested population in present study):

Gold standard pulmonary function test

指标试验（即本研究的待评估诊断试验，无论为方法、生物标志物或设备，均请列出名称）：

人工智能模型

Index test:

Artificial intelligence model

目标人群（可以是某种疾病患者或正常人群，详细描述其疾病特征，注意应纳入符合分布特点的全序列病例，具有良好的代表性）

慢性阻塞性肺病患者

例数:

Sample size:

250

Target condition (The target condition is a particular disease or disease stage that the index test will be intended to identify. Please specify the characteristics in detail; the population should has a complete spectrum and good representative):

Patients with chronic obstructive pulmonary disease

容易混淆的疾病人群（即与目标疾病不易区分的一种或多种不同疾病，应避免采用正常人群对照的病例-对照设计）：

无

例数:

Sample size:

0

Population with condition difficult to distinguish from the target condition, the normal population in a case-control study design should be avoid:

None

研究实施地点：

Countries of recruitment and research settings：

国家：

 中国

省(直辖市)：

 上海 

市(区县)：

  

Country：

China 

Province：

Shanghai 

City：

 

单位(医院)：

海军军医大学长海医院 

单位级别：

三甲 

Institution
hospital：

Changhai Hospital, The Navy Military Medical University

Level of the institution：

Tertiary A

测量指标：

Outcomes：

指标中文名：

AI预测值与肺功能实测值的均值差(FEV1、FVC)

指标类型：

主要指标

Outcome：

Mean difference between AI-predicted values and measured pulmonary function values (FEV1, FVC)

Type：

Primary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

指标中文名：

GOLD分级

指标类型：

次要指标

Outcome：

GOLD classification

Type：

Secondary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

指标中文名：

决策改变率

指标类型：

次要指标

Outcome：

Decision Change Rate

Type：

Secondary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

指标中文名：

流程效率

指标类型：

次要指标

Outcome：

Process efficiency

Type：

Secondary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

指标中文名：

患者体验

指标类型：

次要指标

Outcome：

Patient experience

Type：

Secondary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

指标中文名：

一致性

指标类型：

次要指标

Outcome：

Agreement

Type：

Secondary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

指标中文名：

校准度

指标类型：

次要指标

Outcome：

Calibration

Type：

Secondary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

指标中文名：

相关系数

指标类型：

主要指标

Outcome：

Correlation coefficient

Type：

Primary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

采集人体标本：

Collecting sample(s)
from participants：

标本中文名：

无

组织：

Sample Name：

None

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 岁 years
最大 Max age 90 岁 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）：

1采用中心分层置换区组随机化方法： (1)区组设计：使用动态区块大小（随机选择4或6），以降低分组可预测性。 (2)分配比例：试验组（AI信息可见）与对照组（常规流程）为1:1。 (3)实现方法：由独立统计师使用R软件（如blockrand包）生成不可预测的随机分配序列。 2 分层因素 为确保组间基线均衡，随机化按以下两个因子进行分层： (1)研究中心：用以控制不同中心在患者来源、CT设备及临床实践上的差异。 (2)吸烟包年数：分为“>=10包年”与“<10包年/从不吸烟”两层，此为COPD最重要的风险因素之一。

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

1. Adopt the central stratified block randomization method: (1) Block design: Use a dynamic block size (randomly select 4 or 6) to reduce the predictability of the grouping. (2) Allocation ratio: The experimental group (with AI information visible) and the control group (conventional process) are in a 1:1 ratio. (3) Implementation method: An independent statistician uses R software (such as the blockrand package) to generate an unpredictable random allocation sequence. 2. Stratification factors To ensure baseline balance between groups, randomization is stratified according to the following two factors: (1) Research center: To control the differences in patient sources, CT equipment, and clinical practices among different centers. (2) Number of smoking packs per year: Divided into two layers: ">=110 packs per year" and "<10 packs per year / never smoking". This is one of the most important risk factors for COPD.

是否公开试验完成后的统计结果:

Calculated Results after the Study Completed public access:

公开/Public

盲法：

1盲法类型 本研究采用部分盲法（Single-Blind），即患者、结局评估者（肺功能质控人员）与统计分析人员对分组信息保持盲态，而负责第二阶段决策的临床医师因需根据分组信息查看不同报告，故无法设盲。 2盲法实施流程 (1)对患者设盲：在知情同意过程中，使用统一的中性语言告知研究流程，不强调分组细节。所有受试者均需完成CT和肺功能检查，以最大程度降低因知晓分组而产生的期望效应差异。 (2)对评估者/统计分析者设盲： 1)数据脱敏与编码：所有用于分析的肺功能数据、AI预测值及临床决策数据在数据库中均使用匿名ID标识，不包含分组信息。 2)盲态维护：负责肺功能报告质控的人员、以及进行最终统计分析的统计师，在整个数据清理和分析阶段均无法接触分组变量。数据库中的分组信息在主要分析完成、数据库锁定后方可解盲。 (3)临床医师的非盲态管理：尽管临床医师在第二阶段知晓分组，但研究通过标准化流程控制偏倚： 1)第一阶段决策盲法：医师在第一阶段（初步决策）时，对所有患者信息（包括分组和肺功能/AI结果）均处于盲态。 2)报告格式标准化：AI报告与肺功能检查报告在eCRF系统中以格式统一的界面呈现，仅内容来源不同，避免因界面差异引入偏倚。 3 破盲管理与应急处理 (1)破盲条件：破盲仅适用于与试验干预相关的严重不良事件（如因AI信息严重偏差导致临床决策错误并造成患者伤害），且必须知晓分组信息方能进行医疗处理时。 4盲法质量控制 (1)盲法效果评估：研究结束后，可对设盲的统计分析师和评估者进行问卷调查，评估其是否猜破分组，并使用统计学方法分析猜破情况与结果评价是否存在关联。 (2)数据一致性检查：由数据管理员在盲态下核对分析数据集与原始数据，确保盲法实施过程未引入数据错配或错误。

Blinding：

1 Type of blinding This study adopts partial blinding (Single-Blind), where the patients, outcome assessors (lung function quality control personnel), and statistical analysts remain blinded to the group information, while the clinical physicians responsible for the second-stage decision-making cannot be blinded as they need to view different reports based on the group information. 2 Blinding implementation process (1) Blinding of patients: During the informed consent process, a uniform neutral language is used to inform the research process without emphasizing the details of the group assignment. All subjects must complete CT and lung function tests to minimize the difference in expectation effects caused by knowing the group assignment. (2) Blinding of assessors/statistical analysts: 1) Data desensitization and coding: All lung function data, AI prediction values, and clinical decision data used for analysis are identified by anonymous IDs in the database and do not contain group information. 2) Maintenance of blinding: The personnel responsible for lung function report quality control and the statistician conducting the final statistical analysis cannot access the group variables during the entire data cleaning and analysis process. The group information in the database can only be unblinded after the main analysis is completed and the database is locked. (3) Non-blinding management of clinical physicians: Although the clinical physicians are aware of the group assignment in the second stage, the study controls bias through standardized procedures: 1) Blinding of the first-stage decision-making: During the first stage (initial decision-making), the physician is in a blinded state for all patient information (including group assignment and lung function/AI results). 2) Standardized report format: AI reports and lung function examination reports are presented in a uniformly formatted interface in the eCRF system, with only the content sources being different to avoid introducing bias due to interface differences. 3 Blinding management and emergency handling (1) Conditions for unblinding: Unblinding is only applicable to serious adverse events related to the trial intervention (such as clinical decision errors caused by significant deviations in AI information and resulting in patient harm), and medical treatment can only be performed when the group assignment information is known. 4 Blinding quality control (1) Evaluation of blinding effect: After the study, a questionnaire survey can be conducted for the blinded statistical analysts and assessors to evaluate whether they guessed the group assignment and whether there is an association between the guess and the result evaluation using statistical methods. (2) Data consistency check: The data administrator checks the analysis data set and the original data in a blinded state to ensure that no data mismatch or errors are introduced during the implementation of the blinding process.

试验完成后的统计结果（上传文件）：

Calculated Results after
the Study Completed(upload file)：

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）：

发表文章；预测AI模型和推理代码已在GitHub（https://github.com/CHANGHAI-AILab/NET）上公开提供

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

Published article;The prediction AI models and inference code were made publicly available on GitHub (https://github.com/CHANGHAI-AILab/NET)

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：

数据采集：研究者将病例信息记录于l结构化电子病例记录表表格后双人核对，通过电子数据采集系统进行数据录入与管理。数据管理：有关受试者身份相关的所有信息资料均予以保密，相关资料在相关法律和/或法规允许的范围之外不对外公开。

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

Data collection: The researchers recorded the case information on a structured electronic case record form and then double-checked it. The data were entered and managed through an electronic data collection system. Data management: All information related to the identities of the subjects is kept confidential. Such information will not be disclosed outside the permitted scope as stipulated by relevant laws and/or regulations.

数据与安全监察委员会：

Data and Safety Monitoring Committee：

暂未确定/Not yet

注册人：

Name of Registration：

  2026-01-23 08:28:29