Development and validation of a machine learning model for survival risk stratification after esophageal cancer surgery

Abstract

BackgroundPrediction of prognosis for patients with esophageal cancer(EC) is beneficial for their postoperative clinical decision-making. This study’s goal was to create a dependable machine learning (ML) model for predicting the prognosis of patients with EC after surgery.MethodsThe files of patients with esophageal squamous cell carcinoma (ESCC) of the thoracic segment from China who received radical surgery for EC were analyzed. The data were separated into training and test sets, and prognostic risk variables were identified in the training set using univariate and multifactor COX regression. Based on the screened features, training and validation of five ML models were carried out through nested cross-validation (nCV). The performance of each model was evaluated using Area under the curve (AUC), accuracy(ACC), and F1-Score, and the optimum model was chosen as the final model for risk stratification and survival analysis in order to build a valid model for predicting the prognosis of patients with EC after surgery.ResultsThis study enrolled 810 patients with thoracic ESCC. 6 variables were ultimately included for modeling. Five ML models were trained and validated. The XGBoost model was selected as the optimum for final modeling. The XGBoost model was trained, optimized, and tested (AUC = 0.855; 95% CI, 0.808-0.902). Patients were separated into three risk groups. Statistically significant differences (p < 0.001) were found among all three groups for both the training and test sets.ConclusionsA ML model that was highly practical and reliable for predicting the prognosis of patients with EC after surgery was established, and an application to facilitate clinical utility was developed.

Glycosylation-related genes mediated prognostic signature contribute to prognostic prediction and treatment options in ovarian cancer: based on bulk and single‑cell RNA sequencing data

“…With the proliferation of medical data and the rapid development of technology and artificial intelligence, the use of big data analysis to build survival prediction models has become an important research topic. Machine learning, a subfield of artificial intelligence, can identify patterns and relationships in the data and provide accurate predictions of future events [ 27 , 28 ]. Machine learning methods have been used to construct prognostic models for various malignancies, such as lung, liver, breast, and gastrointestinal cancers [ 29 – 32 ].…”

Prediction of five-year survival among esophageal cancer patients using machine learning

“… 360 patients with ESCC Bald eagle search and least-squares support vector machine No BES-LSSVM had a higher accuracy rate, with 86.538% for the high-age group and 86.495% for the low-age group. Xu et al [ 21 ] Five-year 16 features Yes (Univariate and multivariate regression analysis) clinicopathological characteristics and follow-up data of ESCC patients at the Department of Thoracic Surgery in Northern Jiangsu People's Hospital Gender, age, type of surgery, hypertension, diabetes, smoking, drinking, tumor size, tumor center location, histological grade, PT stage, pN stage, vascular invasion, nerve invasion, pathological types, surgical margins, 810 patients with ESCC Decision tree, RF, SVM, GBM, XG-Boost No The XG-Boost model with (AUC = 0.855; 95% CI, 0.808–0.902) was considered optimal. Zhang et al [ 35 ] Three-year and five-year survival 27 features Yes (LASSO regularization and univariable Cox regression analysis) One single-center database of Sichuan Cancer Hospital Age, sex, Karnofsky performance scale score, tumor length, tumor grade, tumor location, vascular invasion, surgical margin, dissected lymph nodes number, nerve invasion, T stage, N stage, AJCC8th stage, surgical intervention alone, hematocrit, mean platelet volume, neutrophil to lymphocyte ratio, monocytes, eosinophil, direct bilirubin, albumin, aspartate aminotransferase, alkaline phosphatase, sodium, magnesium, fibrinogen, lymphocyte -to- monocytes ratio 2441 ESCC patients R-part, Elastic Net, GBM, RF, GLMboost, and ML-extended CoxPH method No ML-extended CoxPH has a 75.4%, 45.8%, and 26.9% prediction capability for stratifying the low, medium, and high-risk groups for three-year survival.…”

“…As a classification system for malignancy, it can be leveraged for cancer prognosis staging and assess the malignancy based on the tumor size, regional lymph node involvement, and metastasis conditions [ 20 ]. Although the TNM staging system is used for screening the prognosis of EC patients, it is not considered a comprehensive approach for accurate prediction due to the limited features used in this system [ 21 ]. Also, considering the low five-year survival rate of EC and the heterogeneity of these cancer cases concerning pathological characteristics and age, enhancing the survival rate and saving EC patients at advanced stages is a global challenge.…”