Development and Real-World Validation of an Intelligent Interpretation Model for Claudin18.2 Protein Expression in Digestive System Tumors
试验速览
- 阶段
- 不适用
- 状态
- Enrolling By Invitation
- 入组人数
- 2,000
- 试验地点
- 2
- 主要终点
- Area under ROC curve (AUC)
研究概览
简要总结
The investigators plan to develop a deep learning-based automatic interpretation model for Claudin18.2(CLDN18.2) using the institution's and multiple other centers' extensive pathological resources of digestive system adenocarcinomas. This study will not only strictly follow the latest domestic expert consensus and standards, but also aims to address current pain points in manual interpretation. It seeks to provide technical support for standardizing, objectifying, and streamlining CLDN18.2 testing, thereby advancing the application of precision medicine in the diagnosis and treatment of digestive system diseases. The project has clear clinical necessity and broad application prospects.
研究设计
- 研究类型
- Observational
- 观察模型
- Cohort
- 时间视角
- Prospective
入排标准
- 年龄范围
- 18 Years 至 —（Adult, Older Adult）
- 性别
- All
- 接受健康志愿者
- 否
入选标准
- •Age over 18 years.
- •Patients who underwent CLDN18.2 immunohistochemistry and H&E staining.
- •Availability of complete pathology reports and clinical information.
排除标准
- •1.Patients with missing data or specimens not meeting quality control requirements for analysis.
研究组 & 干预措施
Prospective validation dataset
We conducted a prospective validation study to compare the diagnostic performance among pathologists, our pathology foundation model, and pathologist-with-AI-assisted diagnosis. This study was initiated on October 1, 2025 at Nanfang Hospital, Southern Medical University (NFHSMU)
QFSH external validation dataset
2000 slides from 1000 eligible individuals were obtained in the Qianfoshan Hospital (QFSH, Jinan, China) between January 2020 and February 2026, which was used to validate the model.
Randomized controlled trial
We conducted a randomized controlled trial（RCT）to compare the diagnostic performance among pathologists, the pathology foundation model, and pathologist-with-model-assisted diagnosis at Nanfang Hospital of Southern Medical University (NFHSMU).The trial commenced data collection on February 1, 2023 to establish the NFHSMU randomized controlled trial dataset. Following quality control, the slides were randomly allocated (1:1 ratio) into three groups:Pathologist-only group/Model-assisted pathologist group.
结局指标
主要结局
Area under ROC curve (AUC)
时间窗: Diagnostic evaluation will be performed within 1 week when the whole slide images(WSIs) are obtained.
Area under the curve
次要结局
- Specificity(Diagnostic evaluation will be performed within 1 week when the WSIs are obtained)
- Sensitivity(Diagnostic evaluation will be performed within 1 week when the WSIs are obtained)