Machine Learning to Predict Lymph Node Metastasis in T1 Esophageal Squamous Cell Carcinoma: A Multicenter Study
试验速览
- 阶段
- 不适用
- 状态
- 已完成
- 发起方
- 入组人数
- 1,267
- 试验地点
- 1
- 主要终点
- Sub-analysis (ML Model vs. Logistic Model vs. NCCN Guideline)
研究概览
简要总结
Existing models do poorly when it comes to quantifying the risk of Lymph node metastases (LNM). This study generated elastic net regression (ELR), random forest (RF), extreme gradient boosting (XGB), and a combined (ensemble) model of these for LNM in patients with T1 esophageal squamous cell carcinoma.
详细描述
Lymph node metastases (LNM) is a relatively uncommon but possible complication of T1 esophageal squamous cell carcinoma (ESCC). Existing models do poorly when it comes to quantifying this risk. This study aimed to develop a machine learning model for LNM in patients with T1 esophageal squamous cell carcinoma.
Patients with T1 squamous cell carcinoma treated with surgery between January 2010 and September 2021 from 3 institutions were included in this study. Machine-learning models were developed using data on patients' age and sex, depth of tumor invasion, tumor size, tumor location, macroscopic tumor type, lymphatic and vascular invasion, and histologic grade. Elastic net regression (ELR), random forest (RF), extreme gradient boosting (XGB), and a combined (ensemble) model of these was generated. Use Area Under Curve (AUC) to evaluate the predictive ability of the model. The contribution to the model of each factor was calculated. In order to better meet clinical needs, the investigators have designed the model as a user-friendly website.
研究设计
- 研究类型
- Interventional
- 分配方式
- Na
- 干预模型
- Single Group
- 主要目的
- Diagnostic
- 盲法
- None
入排标准
- 性别
- All
- 接受健康志愿者
- 否
入选标准
- •(I) thoracic ESCC
- •(II) no history of concomitant or prior malignancy
- •(III) tumor with pT1 staging
- •(IV) 15 or more lymph nodes examined
排除标准
- •underwent neoadjuvant treatment or endoscopic submucosal dissection before surgery
结局指标
主要结局
Sub-analysis (ML Model vs. Logistic Model vs. NCCN Guideline)
时间窗: 8 weeks
Apply NCCN guidelines and logistic models for prediction, and compare their performance with the model obtained in this study to determine the actual application benefits of the model
Model performance: discrimination
时间窗: 8 weeks
Draw the ROC curve of the model and obtain their AUC values, and select the best prediction model based on the results of the validation set
Variable importance
时间窗: 6 weeks
Calculate the importance level of variables used in the model and sort them, and analyze the reasons for the most important variables
次要结局
未报告次要终点