Breast Cancer Prediction Using Artificial Neural Networks: A Comprehensive Evaluation of Diagnostic Accuracy and Performance
Eshika Jain, Aseem Aneja · 2025
In the last few years, ANN has shown huge promise in predictive healthcare, especially in the diagnosis of breast cancer. This work is therefore devoted to developing and accessing an ANN-based model for the prediction of breast cancer. The dataset contains 569 observations, consisting of radius, texture, perimeter, area, and other measures of smoothness, compactness, and symmetry. These features were used to classify the tumors as either benign or malignant. According to the classification report, an overall accuracy of 96% was achieved by the ANN model. It was supported by an appreciably high precision (96%) and recall (99%) for class 0, which corresponded to a very strong F1-score of 97%. In the case of class 1, malignant, the precision attained was 97, while the recall was as high as 92%, with an F1-score of 95%. The weighted average F 1 -score was 96, hence justifying the model’s strength. The most relevant features that contributed to the accuracy of the model were area mean, perimeter mean, and texture mean. The total dataset contains 32 features that relate to various statistical measures that describe physical characteristics of the tumor. The model's performance underlines the great potential of ANNs in early detection of breast cancer, thereby increasing diagnostic capabilities. The application of such data-driven techniques may enable clinicians to minimize false positives and negatives, which would increase the efficiency of treatment. This study further builds upon the increasing trend of the application of artificial intelligence into clinical use and limits the use of further advanced predictive models in medical diagnostics.