An Integrated Statistical and Machine Learning Approach for Breast Cancer Classification Using Tumor Morphological...

Area of Science:

• Oncology
• Biostatistics
• Machine Learning

Background:

• Breast cancer is a leading cause of death in women globally.
• Accurate tumor classification is crucial for effective patient treatment.
• Integrating statistical and machine learning methods can enhance diagnostic accuracy.

Purpose of the Study:

• To evaluate an integrated statistical and machine learning framework for breast cancer classification.
• To compare the performance of logistic regression, random forest, and support vector machine (SVM) models.
• To assess the potential of these models in supporting early breast cancer detection.

Main Methods:

• Retrospective analysis of the Breast Cancer Wisconsin Diagnostic Dataset (569 samples).
• Utilized eight morphological features and stratified random sampling for data splitting (70% training, 30% testing).
• Developed and evaluated logistic regression, random forest, and SVM models using cross-validation and performance metrics (accuracy, AUC, etc.).

Main Results:

• Morphological features like radius, texture, smoothness, and concavity significantly predicted malignancy.
• Logistic regression yielded the highest accuracy (95.3%) and AUC (0.983) on the test set.
• SVM and random forest models also demonstrated strong, comparable performance, with no significant differences in AUC among models.

Conclusions:

• An integrated statistical and machine learning approach offers a robust method for breast cancer classification.
• Logistic regression slightly outperformed machine learning models, but all achieved high accuracy.
• This combined approach shows significant potential for improving early detection and clinical decision-making in breast cancer diagnosis.