Advancing Breast Cancer Diagnosis: A Comprehensive Machine Learning Approach for Predicting Malignant and Benign Cases with Precision and Insight in a Neutrosophic Environment using Neutrosophic Numbers

Abstract

Breast cancer is still among the deadliest diseases globally, and its detection in an early stage still 
represents a big challenge in medical diagnostics. This research suggests a complete machine 
learning framework to predict the probability of benign and malignant breast cancer cases with 
improved accuracy and interpretability. The work uses an established dataset, and for 
comparative analysis and for insights into the data distribution, statistical analysis is also 
incorporated. Four top machine learning algorithms are trained and evaluated with a series of 
performance measures such as accuracy, positive predictive value (PPV), negative predictive  value (NPV), F1-score, etc. In order to compensate for inherent uncertainties and imprecise in 
clinical data, the paper proposes a neutrosophic logic with neutrosophic numbers for improved 
decision-making. The results show the efficacy of using machine learning with neutrosophic 
theory to enhance diagnostic accuracy and facilitate early intervention measures in the treatment 
of breast cancer.  

DOI: 10.5281/zenodo.15733455