Related Experiment Video
Updated: Aug 13, 2026
Comparison of Predictive Performance of Three Lymph Node Staging Systems in Colorectal Signet Ring Cell Carcinoma Based on Machine Learning Model
Published on: April 18, 2025
Building Machine Learning Models Based on Oxidative Stress Index Score to Predict Survival of Locally Advanced Rectal
Baoqiu Liu1,2,3, Li Zhuo1,4, Yirui Wang1
1Department of Radiation Oncology, State Key Laboratory of Oncology in South China, Collaborative Innovation Center for Cancer Medicine, Guangdong Provincial Clinical Research Center for Cancer, Sun Yat-Sen University Cancer Center, Guangzhou, China.
A new oxidative stress index score, integrating liver enzymes, effectively predicts survival in locally advanced rectal cancer patients. Machine learning models using this score show superior prognostic value over traditional methods.
Area of Science:
- Oncology
- Biomarkers
- Machine Learning
Background:
- Liver enzyme biomarkers are implicated in colorectal cancer development and progression.
- Accurate prognostic tools are crucial for optimizing treatment strategies in locally advanced rectal cancer.
Purpose of the Study:
- To develop and validate a novel oxidative stress index score using gamma-glutamyl transferase and total bilirubin.
- To assess the prognostic value of this score in locally advanced rectal cancer patients undergoing neoadjuvant therapy.
- To construct and validate machine learning-based survival prediction models incorporating the oxidative stress index score.
Main Methods:
- A novel oxidative stress index score was developed by integrating gamma-glutamyl transferase and total bilirubin levels.
- Machine learning models, including random survival forests, were built using the oxidative stress index score and clinicopathological data.
- Model performance was evaluated and validated in training and validation cohorts of locally advanced rectal cancer patients.
Main Results:
- Random survival forests demonstrated the highest predictive efficiency among tested machine learning models.
- The oxidative stress index score-based random survival forests models significantly outperformed ypstage in predicting disease-free and overall survival.
- Patients stratified into low-risk groups showed significantly better survival outcomes compared to high-risk groups in both cohorts.
Conclusions:
- The novel oxidative stress index score is a valuable prognostic indicator in locally advanced rectal cancer.
- Machine learning models incorporating the oxidative stress index score offer superior predictive performance for survival outcomes.
- These findings support the use of advanced predictive models for personalized risk stratification and treatment planning.