An Improved Early Breast Cancer Cells Classification and Prediction Based on a Fuzzy Neural Network Model

Breast cancer is the most common type of cancer among women. Accurately diagnosis requires experienced medical practitioners to determine the nature of the cells. However, given the inherent complexity, there is a potential risk of misdiagnosis. This study proposes an artificial intelligence system that integrates fuzzy reasoning and a neural network to accurately classify cells as benign or malignant. Using the Wisconsin Breast Cancer (Diagnosis) dataset, samples were randomly partitioned into a training set of 400 and a testing set of 169 samples, following a 7:3 ratio. It is worth noting that these samples are correlated with 30 parameters, which can be computationally demanding. To address this issue, the principal component analysis (PCA) technique was employed to eliminate less significant parameters, resulting in a reduced set of only 6 key parameters. The proposed PCA NF model achieved a test accuracy of 97.63%, with 100% precision, 93.10% recall, and a 96.43% F-measure. The PCA-ANFIS model achieved 95.27% accuracy and 94.12% for both precision and recall. Both models demonstrated reliable discrimination, supported by Matthews correlation coefficients of 94.80% and 90.16% for PCA-NF and PCA-ANFIS, respectively. The research novelty lies in the enhanced ANFIS approach, which provides comparable accuracy to existing artificial intelligence techniques while simplifying the diagnosis process. This user-friendly approach greatly benefits clinical medical experts by enhancing workflow efficiency and effectiveness.