An accurate and interpretable model for BCCT.core.

Area of Science:

• Medical imaging analysis
• Machine learning in healthcare
• Breast cancer treatment outcomes

Background:

• Breast Cancer Conservative Treatment (BCCT) is a common breast cancer therapy.
• Objective evaluation of BCCT aesthetic outcomes is challenging due to subjective assessments.
• Current BCCT.core system uses a support vector machine (SVM) with radial basis function (RBF) for aesthetic evaluation, but lacks interpretability.

Purpose of the Study:

• To investigate the accuracy of interpretable machine learning models for evaluating BCCT aesthetic results.
• To compare the performance of decision trees and linear classifiers against the existing RBF SVM in BCCT.core.
• To determine if interpretable models can replace the current RBF SVM without compromising accuracy.

Main Methods:

• Comparison of decision trees and linear classifiers with the RBF SVM classifier.
• Utilizing features computed from digital photographs of patients undergoing BCCT.
• Experimental study to evaluate the predictive accuracy of different models.

Main Results:

• Interpretable models (decision trees, linear classifiers) demonstrated accuracy comparable to the RBF SVM.
• The performance of the BCCT.core system was maintained when using interpretable models.
• Suggests that the current RBF SVM can be replaced by more interpretable alternatives.

Conclusions:

• Interpretable machine learning models are viable alternatives for objective aesthetic evaluation in BCCT.
• Replacing the RBF SVM in BCCT.core with decision trees or linear classifiers is feasible without performance loss.
• This research enhances the interpretability of BCCT aesthetic outcome assessment systems.