ChiCTR2600128491 版本V1.0 版本创建时间2026/07/21 15:43:33 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：

 ChiCTR2600128491 

最近更新日期：

Date of Last Refreshed on：

 2026-07-21 15:43:24 

注册时间：

Date of Registration：

 2026-07-21 00:00:00 

注册号状态：

预注册

Registration Status：

Prospective registration

注册题目：

基于人工智能动态预警系统辅助的ICU患者低氧风险早期识别与干预效果评价：一项多中心、前瞻性、随机、平行对照试验

Public title：

Clinical Value of an AI-Based Dynamic Early Warning System-Assisted Approach for Early Identification and Intervention of Hypoxemia Prediction in ICU Patients: A Multicenter, Prospective, Randomized, Parallel-Controlled Trial

注册题目简写：

English Acronym：

研究课题的正式科学名称：

基于人工智能动态预警系统辅助的ICU患者低氧风险早期识别与干预效果评价：一项多中心、前瞻性、随机、平行对照试验

Scientific title：

Clinical Value of an AI-Based Dynamic Early Warning System-Assisted Approach for Early Identification and Intervention of Hypoxemia Prediction in ICU Patients: A Multicenter, Prospective, Randomized, Parallel-Controlled Trial

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：

张涛 

研究负责人：

万健 

Applicant：

Zhang Tao 

Study leader：

Wan Jian 

申请注册联系人电话：

Applicant telephone：

 +86 21 2050 9000

研究负责人电话：

Study leader's
telephone：

+86 136 2198 1306

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：

 zhangtao@shpdph.cn

研究负责人电子邮件：

Study leader's E-mail：

 wanjian@shpdph.cn

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：

中国上海市浦东新区川沙镇川环南路490号

研究负责人通讯地址：

中国上海市浦东新区川沙镇川环南路490号

Applicant address：

490 Chuanhuan South Road, Chuansha Town, Pudong New Area, Shanghai, China

Study leader's address：

490 Chuanhuan South Road, Chuansha Town, Pudong New Area, Shanghai, 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：

上海市浦东新区人民医院

Applicant's institution：

Shanghai Pudong New Area People's Hospital

研究负责人所在单位：

上海市浦东新区人民医院

Affiliation of the Leader：

Shanghai Pudong New Area People's Hospital

是否获伦理委员会批准：

是

Approved by ethic committee：

Yes

伦理委员会批件文号：

Approved No. of ethic committee：

 (2026)伦审第(K65)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：

上海市浦东新区人民医院伦理委员会

Name of the ethic committee：

Ethics Committee of Shanghai Pudong New Area People's Hospital

伦理委员会批准日期：

Date of approved by ethic committee：

 2026-05-12 00:00:00

伦理委员会联系人：

周雪

Contact Name of the ethic committee：

Zhou Xue

伦理委员会联系地址：

中国上海市浦东新区川沙镇川环南路490号

Contact Address of the ethic committee：

490 Chuanhuan South Road, Chuansha Town, Pudong New Area, Shanghai, China

伦理委员会联系人电话：

Contact phone of the ethic committee：

 +86 131 6226 6991

伦理委员会联系人邮箱：

Contact email of the ethic committee：

 zhouxue@shpdph.cn

研究实施负责（组长）单位：

上海市浦东新区人民医院

Primary sponsor：

Shanghai Pudong New Area People's Hospital

研究实施负责（组长）单位地址：

中国上海市浦东新区川沙镇川环南路490号

Primary sponsor's address：

490 Chuanhuan South Road, Chuansha Town, Pudong New Area, Shanghai, 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)：

上海

市(区县)：

浦东新区

Country：

China

Province：

Shanghai

City：

Pudong New Area

单位(医院)：

上海市浦东新区人民医院

具体地址：

中国上海市浦东新区川沙镇川环南路490号

Institution
hospital：

Shanghai Pudong New Area People's Hospital

Address：

490 Chuanhuan South Road, Chuansha Town, Pudong New Area, Shanghai, China

经费或物资来源：

上海市经济和信息化委员会

Source(s) of funding：

Shanghai Municipal Commission of Economy and Informatization

研究疾病：

低氧血症  

Target disease：

Hypoxemia

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：

 探索性研究/预试验 

Study phase：

0

研究设计：

随机平行对照 

Study design：

Parallel 

研究目的：

主要目的：评价医护人员在常规ICU监护基础上使用人工智能低氧风险动态预警系统进行辅助管理，是否能较常规医护人员管理降低ICU患者低氧事件发生率。 次要目的： 1.比较AI辅助医护人员管理与常规医护人员管理在低氧事件早期识别方面的差异：包括有效预警率和预警提前时间（从首次预警到低氧发生的时间间隔）； 2.比较两组患者低氧事件严重程度：包括低氧持续时间及最低SpO?水平； 3.比较两组医护人员临床干预时机：包括吸氧、无创通气或机械通气等干预启动时间； 4.比较两组患者临床结局：包括ICU住院时间、机械通气使用率及ICU死亡率； 5.评估AI系统对医护人员临床决策过程的影响：包括管理建议改变率及医护人员对预警信息的采纳率； 6.评估AI辅助系统在不同患者亚组中的应用效果：如呼吸系统疾病患者、术后患者及不同病情严重程度人群。  

Objectives of Study：

Primary Objective To evaluate whether the application of an artificial intelligence-based dynamic early warning system for hypoxia risk as an adjunct to routine ICU monitoring and management can reduce the incidence of hypoxic events in ICU patients compared with conventional management. Secondary Objectives 1.To compare the performance of AI-assisted management and conventional management in the early identification of hypoxic events, including effective warning rate and advance warning time (the time interval from the first warning to the onset of hypoxia). 2.To compare the severity of hypoxic events between the two groups, including the duration of hypoxia and the minimum SpO? level. 3.To compare the timing of clinical interventions implemented by medical staff in the two groups, including the initiation time of oxygen therapy, non-invasive ventilation, mechanical ventilation and other interventions. 4.To compare clinical outcomes of patients in the two groups, including length of ICU stay, rate of mechanical ventilation and ICU mortality. 5.To assess the impact of the AI system on the clinical decision-making process of medical staff, including the rate of revised management plans and the acceptance rate of warning information among medical staff. 6.To evaluate the efficacy of the AI-assisted system in different patient subgroups, such as patients with respiratory diseases, postoperative patients, and populations with different disease severity.

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，性别不限； 2.入住ICU并接受连续生命体征监测（包括SpO?、心率、呼吸频率、血压等）； 3.预计ICU停留时间 >=24 小时，具备连续观察低氧风险的条件； 4.入组时未发生持续性低氧（SpO? >=93%或经简单氧疗可维持）； 5.生命体征监测数据质量达标，可用于AI系统实时分析； 6.受试者本人或其法定代理人签署知情同意书。

Inclusion criteria

1.Aged 18 years or older, with no restriction on gender. 2.Admitted to the ICU and receiving continuous vital sign monitoring (including SpO?, heart rate, respiratory rate, blood pressure, etc.). 3.Expected ICU stay of no less than 24 hours, allowing continuous monitoring of hypoxia risk. 4.No persistent hypoxia at enrollment (SpO? >= 93% or stable with simple oxygen therapy). 5.Qualified vital sign monitoring data available for real-time analysis by the AI system. 6.Written informed consent obtained from the patient or their legal representative.

排除标准：

1.入组时已存在持续性严重低氧（如SpO? < 93%且持续存在或需立即呼吸支持）； 2.已接受稳定机械通气且短期内通气策略不具备调整空间； 3.预计ICU停留时间 <24 小时或即将转出ICU； 4.终末期疾病或已决定放弃积极治疗的患者； 5.存在严重数据缺失或监测设备无法稳定获取生命体征数据； 6.参与其他可能影响本研究主要终点的干预性临床试验； 7.研究者认为不适合纳入本研究的其他情况。

Exclusion criteria：

1.Presence of persistent severe hypoxia at enrollment (e.g., sustained SpO? < 93% or requiring immediate respiratory support). 2.Receiving stable mechanical ventilation with no room for ventilator adjustment in the short term. 3.Expected ICU stay of less than 24 hours or pending ICU transfer. 4.Patients with end-stage diseases or for whom active treatment has been discontinued. 5.Severe missing data or unstable vital sign acquisition via monitoring devices. 6.Participation in other interventional clinical trials that may affect the primary endpoint of this study. 7.Other conditions deemed inappropriate for enrollment by the investigators.

研究实施时间：

Study execute time：

从 From 2025-12-16 00:00:00至 To 2027-12-31 00:00:00  

征募观察对象时间：

Recruiting time：

从 From 2026-08-01 00:00:00 至 To 2027-06-30 00:00:00

干预措施：

Interventions：

组别：

实验组

样本量：

 430

Group：

Experimental group

Sample size：

干预措施：

在常规 ICU 监护与管理基础上，引入人工智能低氧风险动态预警系统进行辅助管理。系统基于多参数生命体征数据的实时采集与时序建模分析，输出当前低氧风险等级、预警触发时间及风险变化趋势等信息，临床医护人员结合系统预警结果与临床判断制定最终干预决策。AI系统不自动触发治疗措施，也不替代医护人员作出诊疗决策。医护人员可根据患者病情、常规监测信息及AI预警信息综合判断是否采取干预措施，并记录是否采纳AI提示及其对管理策略的影响。

干预措施代码：

Intervention：

On the basis of routine ICU monitoring and management, an artificial intelligence-based dynamic early warning system for hypoxemia risk was introduced for auxiliary management. The system performs real-time collection and time-series modeling analysis of multi-parameter vital sign data, and outputs information including the current hypoxemia risk level, early warning trigger time, and risk change trend. Clinical medical staff formulate final intervention decisions by combining the system’s early warning results with clinical judgment. The AI system will not automatically initiate therapeutic measures, nor replace medical staff in making diagnosis and treatment decisions. Medical staff may comprehensively determine whether to implement interventions based on patients’ conditions, routine monitoring data and AI early warning information, and record whether the AI prompts are adopted as well as their impacts on management strategies.

Intervention code：

组别：

对照组

样本量：

 430

Group：

Control group

Sample size：

干预措施：

由具有 ICU 临床工作经验且未参与本研究人工智能模型开发与训练的临床医护人员负责患者监护与管理，严格遵循 ICU 相关诊疗规范及标准流程（如呼吸支持与氧疗管理规范）开展标准化监护。

干预措施代码：

Intervention：

Patient monitoring and management were performed by clinical medical staff with ICU clinical experience, who were not involved in the development and training of the artificial intelligence model for this study. Standardized monitoring was implemented in strict accordance with relevant ICU diagnosis and treatment specifications and standard operating procedures (e.g., protocols for respiratory support and oxygen therapy management).

Intervention code：

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)：

 浦东新区 

Country：

China

Province：

Shanghai

City：

Pudong New Area

单位(医院)：

 上海市浦东新区人民医院 

单位级别：

 三级 

Institution
hospital：

Shanghai Pudong New Area People's Hospital

Level of the institution：

Tertiary

国家：

 中国

省(直辖市)：

 上海 

市(区县)：

  

Country：

China

Province：

Shanghai

City：

单位(医院)：

 上海市浦东医院 

单位级别：

 三级 

Institution
hospital：

Shanghai Pudong Hospital

Level of the institution：

Tertiary

国家：

 中国

省(直辖市)：

 上海 

市(区县)：

  

Country：

China

Province：

Shanghai

City：

单位(医院)：

 上海市闵行区中心医院 

单位级别：

 三级 

Institution
hospital：

Shanghai Minhang District Central Hospital

Level of the institution：

Tertiary

测量指标：

Outcomes：

指标中文名：

氧饱和度

指标类型：

主要指标

Outcome：

SpO2

Type：

Primary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

指标中文名：

呼吸频率

指标类型：

主要指标

Outcome：

Respiratory rate

Type：

Primary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

指标中文名：

心率

指标类型：

主要指标

Outcome：

Heart rate

Type：

Primary indicator

测量时间点：

测量方法：

Measure time point of outcome：

Measure method：

指标中文名：

血压

指标类型：

主要指标

Outcome：

Blood pressure

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：

正在进行

Recruiting

年龄范围：

Participant age：
最小 Min age 18 岁 years
最大 Max age NA 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）：

本研究采用分层区组随机方法进行分组。考虑到本研究为多中心设计，不同研究中心之间在受试者来源、诊疗流程及管理条件方面可能存在一定差异，拟以研究中心作为主要分层因素，在各中心内按照1:1比例将受试者随机分配至试验组和对照组。 随机序列由独立统计人员采用计算机随机数生成程序制定，并使用区组设计（区块大小4/6），以保证随机分配的均衡性和不可预测性。随机分配方案通过中心随机系统实施，确保受试者分组过程规范、有序。

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

This study adopts stratified block randomization for group allocation. Given the multicenter design, study sites may differ in patient sources, clinical workflows and management conditions. Therefore, study site is set as the primary stratification factor. Within each site, participants are randomly assigned to the experimental group and the control group at a 1:1 ratio. The randomization sequence is generated by an independent statistician using a computer random number generator with block sizes of 4 and 6, to ensure balanced and unpredictable allocation. The randomization scheme is implemented via a central randomization system to guarantee standardized and orderly group assignment.

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：

本研究采用开放标签设计。由于干预措施涉及动态风险评估系统在临床管理中的实际应用，受试者及临床医生不设盲。为尽可能降低评估偏倚，结局评估人员及数据分析人员采用盲法，即在不知晓受试者分组信息的情况下完成结局判定与统计分析。

Blinding：

This is an open-label study. Since the intervention involves the clinical application of a dynamic risk assessment system, blinding of participants and clinicians is not feasible. To minimize assessment bias, outcome assessors and data analysts are kept blinded, who perform outcome evaluation and statistical analysis without access to group allocation information.

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）：

否

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：

本研究将采用统一设计的病例报告表（CRF）和电子数据采集系统（EDC）/电子数据库进行研究数据记录与管理。CRF内容涵盖受试者基线资料、随机分组信息、连续生命体征监测数据、AI低氧风险预警结果、临床干预措施、主要及次要结局指标、不良事件及方案偏离等内容，以确保研究数据采集的完整性、规范性和可追溯性。 基线阶段主要收集受试者人口学信息、既往病史、基础疾病情况及初始生命体征（如SpO?、心率、呼吸频率、血压等）；试验过程中记录随机分组信息、常规监护数据、临床干预措施（如吸氧、无创或有创通气）及相关决策时间；对于试验组，还需记录AI系统输出结果，包括低氧风险评分、预警触发时间、预警次数、医护人员是否采纳AI预警及是否改变原有处理策略；在观察阶段，持续收集低氧事件相关数据（包括发生时间、持续时间及最低SpO?水平）、临床干预措施及ICU结局（如住院时间、机械通气使用情况及死亡情况），用于主要及次要终点的评估。所有数据均由经过统一培训的研究人员按照预设时间点及标准进行采集与录入。 电子数据录入工具应具有用户权限管理、数据修改留痕、逻辑校验、缺失值提示和数据备份等功能，以提高数据录入的可靠性和准确性。对于关键变量，如如随机分组、主要结局指标、安全性数据等，将进行重点核查；必要时采用双人核对或源数据核查方式，确保录入数据与原始资料一致。研究过程中形成的纸质和电子资料均应妥善保存，以保证数据真实、完整、准确和可追溯。

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

This study uses uniformly designed Case Report Forms (CRF) and an Electronic Data Capture (EDC) system/electronic database for data documentation and management. The CRF includes participants’ baseline information, randomization details, continuous vital sign monitoring data, AI-generated hypoxia risk alerts, clinical interventions, primary and secondary outcome measures, adverse events and protocol deviations, so as to ensure the completeness, standardization and traceability of collected data. During the baseline period, demographic data, medical history, underlying diseases and initial vital signs (including SpO?, heart rate, respiratory rate and blood pressure) are collected. During the trial, randomization information, routine monitoring data, clinical interventions (e.g., oxygen therapy, non-invasive and invasive ventilation) and the corresponding decision-making time points are recorded. For the experimental group, outputs of the AI system are additionally documented, including hypoxia risk scores, alert trigger time, number of alerts, as well as whether medical staff adopted the AI alerts and adjusted existing management strategies. Throughout the observation period, data related to hypoxic events (onset time, duration and minimum SpO? level), clinical interventions and ICU outcomes (length of ICU stay, use of mechanical ventilation and mortality) are continuously collected for the assessment of primary and secondary endpoints. All data are collected and entered by uniformly trained researchers in accordance with scheduled time points and standard procedures. The EDC system is equipped with functions such as user privilege management, audit trail for data modifications, logical validation, missing value reminders and data backup to enhance the reliability and accuracy of data entry. Key variables including group allocation, primary outcomes and safety data will undergo intensive verification. Double data entry or source data verification will be performed when necessary to guarantee consistency between entered data and source documents. All paper and electronic documents generated during the study shall be properly retained to ensure the authenticity, integrity, accuracy and traceability of research data.

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：

  2026-07-21 15:43:24