A Novel Approach for Predicting the Survival of Colorectal Cancer Patients Using Machine Learning Techniques and...

Area of Science:

• Oncology
• Biostatistics
• Machine Learning

Background:

• Colorectal cancer (CRC) is a leading cause of cancer mortality globally.
• There is a concerning rise in CRC diagnoses among adults under 50.
• Advanced technologies like artificial intelligence (AI) are crucial for improving CRC outcomes.

Purpose of the Study:

• To develop and evaluate AI-based classification models for predicting colorectal cancer patient survival.
• To assess the performance of various machine learning algorithms in forecasting mortality.
• To identify effective AI tools for clinical decision support in oncology.

Main Methods:

• Eight machine learning classifiers were employed: Random Forest, XGBoost, CatBoost, LightGBM, Gradient Boosting, Extra Trees, k-nearest neighbor (KNN), and decision trees.
• Algorithm optimization was performed using Optuna, RayTune, and HyperOpt frameworks.
• The study utilized a large public dataset from Brazil comprising tens of thousands of patient records.

Main Results:

• The developed models achieved high accuracy in predicting one-, three-, and five-year survival rates.
• Models accurately forecasted overall mortality and cancer-specific mortality.
• Top-performing classifiers (CatBoost, LightGBM, Gradient Boosting, Random Forest) reached approximately 80% accuracy.

Conclusions:

• Effective AI classification models for colorectal cancer prognosis have been developed.
• These models demonstrate potential for integration into clinical practice.
• The research supports the use of AI for enhanced colorectal cancer patient management.