Predicting Characteristics Associated with Breast Cancer Survival Using Multiple Machine Learning Approaches

Area of Science:

• Oncology
• Biostatistics
• Machine Learning in Healthcare

Background:

• Breast cancer is a leading global female cancer.
• Existing survival prediction studies often rely on basic statistical methods.
• There is a need for advanced analytical techniques to identify prognostic markers.

Purpose of the Study:

• To develop and evaluate machine learning models for predicting breast cancer survival.
• To identify key prognostic indicators influencing breast cancer survival rates.
• To assess the utility of machine learning as an alternative prognostic tool.

Main Methods:

• Utilized a hospital-based breast cancer dataset from the SEER Program (2006-2010).
• Included patients with infiltrating duct and lobular carcinoma.
• Constructed prediction models using K-nearest neighbor (K-NN), decision tree (DT), gradient boosting (GB), random forest (RF), AdaBoost, logistic regression (LR), voting classifier, and support vector machine (SVM).

Main Results:

• All machine learning methods demonstrated high accuracy and calibration.
• Logistic regression yielded the lowest accuracy (80.57%), while random forest achieved the highest (94.64%).
• The models effectively identified important prognostic markers for breast cancer survival.

Conclusions:

• Machine learning algorithms are effective tools for predicting breast cancer survival.
• Random forest demonstrated superior performance among the tested models.
• These advanced approaches can serve as valuable alternative prognostic tools, especially in Asian populations.