基于人工智能与多模态数据融合的牙周病精准诊断系统研发与临床验证

注册号:

Registration number:

ChiCTR2600126736 

最近更新日期:

Date of Last Refreshed on:

2026-06-15 11:23:32 

注册时间:

Date of Registration:

2026-06-15 00:00:00 

注册号状态:

预注册

Registration Status:

Prospective registration

注册题目:

基于人工智能与多模态数据融合的牙周病精准诊断系统研发与临床验证

Public title:

Research on the development and clinical validation of an AI-driven precision diagnostic system for periodontal disease with multimodal data fusion

注册题目简写:

English Acronym:

研究课题的正式科学名称:

基于人工智能与多模态数据融合的牙周病精准诊断系统研发与临床验证

Scientific title:

Research on the development and clinical validation of an AI-driven precision diagnostic system for periodontal disease with multimodal data fusion

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:

胡文杰 

研究负责人:

胡文杰 

Applicant:

Hu Wenjie 

Study leader:

Wenjie Hu 

申请注册联系人电话:

Applicant telephone:

+86 10 82195368

研究负责人电话:

Study leader's
telephone:

+86 10 82195368

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:

huwenjie@pkuss.bjmu.edu.cn

研究负责人电子邮件:

Study leader's E-mail:

huwenjie@pkuss.bjmu.edu.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:

北京大学口腔医学院·口腔医院牙周科, 北京 100081

研究负责人通讯地址:

北京市海淀区中关村南大街22号

Applicant address:

100081 Department of Periodontology, Peking University School and Hospital of Stomatology, Beijing 1

Study leader's address:

No.22, Zhongguancun South Avenue,Haidian District, Beijing, 100081

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:

北京大学口腔医院

Applicant's institution:

Peking University School of Stomatology

研究负责人所在单位:

北京大学口腔医院

Affiliation of the Leader:

Peking University Hospital of Stomatology

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:

PKUSSIRB-2026121041

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:

北京大学口腔医院生物医学伦理委员会

Name of the ethic committee:

IRB of Peking University Hospital of Stomatology

伦理委员会批准日期:

Date of approved by ethic committee:

2026-05-27 00:00:00

伦理委员会联系人:

贾效伟

Contact Name of the ethic committee:

jia xiaowei

伦理委员会联系地址:

北京市海淀区中关村南大街22号

Contact Address of the ethic committee:

No.22, Zhongguancun South Avenue,Haidian District, Beijing, 100081

伦理委员会联系人电话:

Contact phone of the ethic committee:

+86 10 82195759

伦理委员会联系人邮箱:

Contact email of the ethic committee:

keyanchuethics@163.com

研究实施负责(组长)单位:

北京大学口腔医院

Primary sponsor:

Peking University Hospital of Stomatology

研究实施负责(组长)单位地址:

北京市海淀区中关村南大街22号

Primary sponsor's address:

No.22, Zhongguancun South Avenue,Haidian District, Beijing, 100081

试验主办单位(项目批准或申办者):

Secondary sponsor:

国家:

中国

省(直辖市):

北京市

市(区县):

Country:

China

Province:

Beijing

City:

单位(医院):

北京大学口腔医院

具体地址:

北京市海淀区中关村南大街22号

Institution
hospital:

Peking University Hospital of Stomatology

Address:

No.22, Zhongguancun South Avenue,Haidian District, Beijing, 100081

经费或物资来源:

首都卫生发展科研专项

Source(s) of funding:

Capital Health Development Research Special Project

研究疾病:

牙周病  

Target disease:

Periodontal disease

研究疾病代码:

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:

基于已构建的多模态大模型,进一步融合临床、影像学指标、关键生物标志物和引入可学习的门控机制,实现三模态融合,并通过监督微调实现迁移学习,建立科学、全面的牙周病诊断分型指标体系。开发具有高度临床实用性的牙周病人工智能多模态融合诊断模型,实现牙周病自动识别与精准分期分级。在外部数据集验证诊断模型,完成真实世界验证。  

Objectives of Study:

Based on the established multimodal large model, we further integrate clinical indicators, imaging parameters and key biomarkers, and introduce learnable gating mechanisms to achieve trimodal fusion. Through supervised fine-tuning for transfer learning, a scientific and comprehensive indicator system for the diagnosis and classification of periodontal diseases is constructed. We develop a clinically practical AI multimodal fusion diagnostic model for periodontal diseases to realize automatic identification and accurate staging and grading of periodontal lesions. The model is validated on external datasets and verified in real-world clinical settings.

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.确认为乙肝、结核、丙肝、梅毒阳性、HIV病史等;
2.妊娠期或哺乳期妇女;
3.近6个月内进行过牙周治疗(牙周基础治疗、手术治疗);
4.近3个月内服用特殊药物(如抗凝血药、激素、抗生素等药物);
5.接受过正畸或正颌治疗;
6.双磷酸盐用药史及头颈部放疗史;

Exclusion criteria:

1.Confirmed history of hepatitis B, tuberculosis, hepatitis C, syphilis, HIV infection, etc.
2.Pregnant or breastfeeding women.
3.Received periodontal treatment (basic periodontal therapy or surgical treatment) within the previous 6 months.
4.Taken specific medications (e.g., anticoagulants, hormones, antibiotics) within the previous 3 months.
5.Received orthodontic or orthognathic surgery.
6.History of bisphosphonate use or head and neck radiotherapy.

研究实施时间:

Study execute time:

From 2026-05-01 00:00:00 To 2028-12-31 00:00:00  

征募观察对象时间:

Recruiting time:

From 2026-06-15 00:00:00 To 2028-04-30 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):

以6名副高级及以上职称牙周病学专家依据2018年美国/欧洲牙周病学会牙周病国际新分类,联合牙周临床检查+ 影像学检查作出的诊断结果作为金标准。

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):

The diagnostic results made by six experts with associate senior professional titles or above in periodontology, based on the new international classification of periodontal diseases released by the American and European Periodontal Societies in 2018, combined with clinical periodontal examinations

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):

基于人工智能与多模态数据融合的牙周病精准诊断系统

Index test:

Precise Diagnosis System for Periodontal Diseases Based on Artificial Intelligence and Multimodal Data Fusion

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)

①年龄18~60周岁,性别不限②符合2018年牙周病新分类诊断标准③牙周健康对照者④能够配合完成检查、样本采集与知情同意。

例数:

Sample size:

629

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):

① Aged between 18-60 years old, no gender restriction.② Comply with the new classification and diagnostic criteria for periodontal diseases issued in 2018.③ Healthy controls with normal periodontal conditions.④ Able to cooperate with examinations, sample collection and sign the informed consent.

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):

非菌斑性牙龈炎(遗传性龈纤维瘤病等)、创伤性牙龈损伤、龈乳头炎等。

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:

Non-plaque-induced gingivitis (such as hereditary gingival fibromatosis), traumatic gingival injury, papillary gingivitis, etc.

研究实施地点:

Countries of recruitment and research settings:

国家:

中国

省(直辖市):

北京市 

市(区县):

 

Country:

China

Province:

Beijing

City:

单位(医院):

北京大学口腔医院 

单位级别:

三级甲等 

Institution
hospital:

Peking University Hospital of Stomatology

Level of the institution:

Tertiary A

国家:

中国

省(直辖市):

北京市 

市(区县):

 

Country:

China

Province:

Beijing

City:

单位(医院):

首都医科大学附属北京朝阳医院 

单位级别:

三级甲等 

Institution
hospital:

Beijing Chao-Yang Hospital Affiliated to Capital Medical University

Level of the institution:

Tertiary A

国家:

中国

省(直辖市):

北京市 

市(区县):

 

Country:

China

Province:

Beijing

City:

单位(医院):

首都医科大学附属北京口腔医院 

单位级别:

三级甲等 

Institution
hospital:

Beijing Stomatological Hospital Affiliated to Capital Medical University

Level of the institution:

Tertiary A

国家:

中国

省(直辖市):

北京市 

市(区县):

 

Country:

China

Province:

Beijing

City:

单位(医院):

首都医科大学宣武医院 

单位级别:

三级甲等 

Institution
hospital:

Xuanwu Hospital, Capital Medical University

Level of the institution:

Tertiary A

测量指标:

Outcomes:

指标中文名:

探诊深度

指标类型:

主要指标

Outcome:

Probing Depth, PD

Type:

Primary indicator

测量时间点:

基线(入组时)

测量方法:

牙周 Williams 刻度探针,精确至 1mm

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

Measured with a Williams periodontal probe, recorded to the nearest 1 mm

指标中文名:

临床附着丧失

指标类型:

次要指标

Outcome:

Clinical attachment loss,CAL

Type:

Secondary indicator

测量时间点:

基线(入组时)

测量方法:

采用 Williams 刻度牙周探针检查,精确至 1mm

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

Measured using a Williams periodontal probe, recorded to the nearest 1 mm

指标中文名:

菌斑指数

指标类型:

次要指标

Outcome:

Plaque Index, PLI

Type:

Secondary indicator

测量时间点:

基线(入组时)

测量方法:

Silness-Loe 标准,0–3 级分级

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

Assessed according to Silness and Loe criteria, graded 0–3

指标中文名:

生物标志物

指标类型:

次要指标

Outcome:

Biomarkers

Type:

Secondary indicator

测量时间点:

基线(入组时)

测量方法:

非刺激性全唾液 3ml,纸尖法采集龈沟液,ELISA 定量检测

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

3 mL unstimulated whole saliva,GCF collected by paper point method, quantified by ELISA

指标中文名:

出血指数

指标类型:

次要指标

Outcome:

Bleeding Index, BI

Type:

Secondary indicator

测量时间点:

基线(入组时)

测量方法:

Mazza 标准,0–5 级分级

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

Assessed according to Mazza criteria, graded 0–5

指标中文名:

牙槽骨丧失高度

指标类型:

次要指标

Outcome:

Alveolar Bone Loss Height

Type:

Secondary indicator

测量时间点:

基线(入组时)

测量方法:

釉牙骨质界至骨嵴顶距离,mm

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

Distance from cementoenamel junction to alveolar crest, in mm

指标中文名:

牙槽骨丧失比例

指标类型:

次要指标

Outcome:

Alveolar bone loss percentage

Type:

Secondary indicator

测量时间点:

基线(入组时)

测量方法:

骨丧失高度占牙根长度百分比

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

Percentage of bone loss relative to root length

指标中文名:

骨缺损形态分类

指标类型:

次要指标

Outcome:

Bone Defect Morphology

Type:

Secondary indicator

测量时间点:

基线(入组时)

测量方法:

水平型 / 垂直型 / 混合型骨吸收

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

Horizontal / Vertical / Combined bone resorption

指标中文名:

骨缺损宽度

指标类型:

次要指标

Outcome:

Bone Defect Width

Type:

Secondary indicator

测量时间点:

基线(入组时)

测量方法:

影像学测量,mm

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

Radiographic measurement, in mm

指标中文名:

骨缺损深度

指标类型:

次要指标

Outcome:

Bone Defect Depth

Type:

Secondary indicator

测量时间点:

基线(入组时)

测量方法:

影像学测量,mm

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

Radiographic measurement, in mm

指标中文名:

骨缺损角度

指标类型:

次要指标

Outcome:

Bone Defect Angle

Type:

Secondary indicator

测量时间点:

基线(入组时)

测量方法:

影像学角度测量,°

Measure time point of outcome:

Baseline (at enrollment)

Measure method:

Radiographic angular measurement, in degrees

指标中文名:

准确率

指标类型:

次要指标

Outcome:

Accuracy

Type:

Secondary indicator

测量时间点:

模型内部验证阶段

测量方法:

模型正确预测的样本数与总样本数的比值:(TP+TN)/(TP+TN+FP+FN)

Measure time point of outcome:

At model internal validation

Measure method:

The ratio of the number of samples correctly predicted by the model to the total number of samples:(TP+TN)/(TP+TN+FP+FN)

指标中文名:

精确率

指标类型:

次要指标

Outcome:

Precision

Type:

Secondary indicator

测量时间点:

模型内部验证阶段

测量方法:

模型预测为正类的样本中实际为正类的比例:TP/(TP+FP)

Measure time point of outcome:

At model internal validatio

Measure method:

The proportion of samples predicted as positive by the model that are actually positive:TP/(TP+FP)

指标中文名:

召回率

指标类型:

次要指标

Outcome:

Recall

Type:

Secondary indicator

测量时间点:

模型内部验证阶段

测量方法:

实际为正类的样本中模型正确预测为正类的比例:TP/(TP+FN)

Measure time point of outcome:

At model internal validation

Measure method:

The proportion of samples that are actually positive classes and correctly predicted as positive by the model:TP/(TP+FN)

指标中文名:

F1 值

指标类型:

次要指标

Outcome:

F1?Score

Type:

Secondary indicator

测量时间点:

模型内部验证阶段

测量方法:

精确率和召回率的调和平均数

Measure time point of outcome:

At model internal validation

Measure method:

The harmonic mean of precision and recall

指标中文名:

ROC 曲线下面积

指标类型:

次要指标

Outcome:

ROC-AUC

Type:

Secondary indicator

测量时间点:

模型内部验证阶段

测量方法:

ROC曲线是以假正例率为横轴,真正例率为纵轴绘制的曲线。AUC是ROC曲线下的面积,其值介于0和1之间。

Measure time point of outcome:

At model internal validation

Measure method:

The ROC curve is plotted with the false positive rate on the horizontal axis and the true positive rate on the vertical axis. AUC refers to the area under the ROC curve, and its value ranges from 0 to 1.

指标中文名:

敏感性

指标类型:

主要指标

Outcome:

Sensitivity

Type:

Primary indicator

测量时间点:

模型外部验证阶段

测量方法:

模型正确识别出实际患病的个体的比例

Measure time point of outcome:

At model internal validation

Measure method:

The proportion of individuals with actual illnesses correctly identified by the model

指标中文名:

特异性

指标类型:

主要指标

Outcome:

Specificity

Type:

Primary indicator

测量时间点:

模型外部验证阶段

测量方法:

模型正确识别出实际未患病的个体的比例

Measure time point of outcome:

At model internal validation

Measure method:

The proportion of individuals who are actually disease-free and correctly identified by the model

指标中文名:

软件易用性

指标类型:

次要指标

Outcome:

Software usability

Type:

Secondary indicator

测量时间点:

模型外部验证阶段

测量方法:

通过测试者的主观感受来评价,采用李克特量表进行量化

Measure time point of outcome:

At model internal validation

Measure method:

Evaluate based on the participants' subjective perceptions and quantify the results using the Likert scale.

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:

模型外部验证阶段

测量方法:

从输入患者数据到模型输出诊断结果所需的时间

Measure time point of outcome:

At model internal validation

Measure method:

The time required from inputting patient data to the model outputting diagnosis results

采集人体标本:

Collecting sample(s)
from participants:

标本中文名:

唾液

组织:

Sample Name:

Saliva

Tissue:

人体标本去向

使用后保存  

说明

Fate of sample:

Preservation after use  

Note:

标本中文名:

龈沟液

组织:

Sample Name:

Gingival crevicular fluid

Tissue:

人体标本去向

使用后保存  

说明

Fate of sample:

Preservation after use  

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 60 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):

学术论文发表后获取通讯作者同意可获取原始数据

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

After the academic paper is published, the original data can be obtained with the consent of the corresponding author

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:

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:

Electronic Capture Management System

数据与安全监察委员会:

Data and Safety Monitoring Committee:

无/No

注册人:

Name of Registration:

 2026-06-15 11:23:32