Early Breast Cancer Prediction using Machine Learning and Deep Learning Techniques

Aug 3, 2026

Breast cancer remains a highly lethal disease worldwide with substantial morbidity and mortality, making accurate early prediction a key requirement for effective diagnosis and treatment. The study compares multiple machine learning classifiers—SVM, KNN, Naïve Bayes, Logistic Regression, Random Forest, Decision Tree, and XGB Classifier—along with a deep learning Artificial Neural Network (ANN) model. Feature-selection and correlation-based selection identify the most informative database attributes. The ANN is implemented with one input layer, two hidden layers, and one output layer. Experimental results show SVM achieves the highest accuracy at about 98.24%, supporting reliable early detection.

LanguageEnglish

CategoryHealthcare

FormatPDF

Pages7

Reading Time18 min

AccessFree to download

I. Introduction

II. Related Works

III. Methodology and Models

IV. Experimental Results

V. Conclusion

VI. References

Early Breast Cancer Prediction using Machine Learning and Deep Learning Techniques 1Ms. Swati B. Patil, Dr. Ritesh V. Patil2, Dr. Parikshit N. Mahalle3 1Department of Computer Engineering Vishawakarma Institute of Information Technology Pune, India [swati.patil@viit.ac.in](swati.patil@viit.ac.in) 2Department of Computer Engineering PEDA’s College Of Engineering, Manjari(BK) Research Guide Department of Computer Engineering Vishawakarma Institute of Information Technology Pune, India [rvpatil3475@yahoo.com](rvpatil3475@yahoo.com) 3Department of Computer Engineering Vishawakarma Institute of Information Technology Pune, India [parikshit.mahalle@viit.ac.in](parikshit.mahalle@viit.ac.in) Abstract: Breast Cancer (BC) is a considered as one of the utmost lethal diseases across the globe that has a very high morbidity and mortality rate. Accurate and early prediction along with diagnosis is one of the most crucial characteristics for the treatment of Breast Cancer. Doctors can have an edge over Breast cancer if they are able to predict it in its early stages using deep learning and machine learning techniques. This paper proposed consists of comparison between the and accuracy of various machine learning models like Support vector machine (SVM), KNearest Neighbours (KNN), Naïve Bayes (NB), Logistic Regression (LR), Random Forest (RF), Decision Tree (DT), XGB Classifier and deep learning model of Artificial neural networks (ANN) for the precise detection of breast cancer. The most crucial properties from the database have been chosen using one feature-selection technique. Correlation is also used to choose the most correlated features from the data. Implementing the ANN model consists of one input layer, two hidden layers, and one output layer. All Machine Learning models and ANN model are then applied to selected features. The results demonstrated that the SVM classifier achieved the highest performance with an accuracy of ~98.24% . Keywords: Breast Cancer, Machine Learning, XGB Classifier, Decision Tree, Naïve Bayes, Logistic Regression, Support Vector Machine, Random Forest, Deep Learning, Artificial Neural Network. I. INTRODUCTION The most common malignant tumour, Breast Cancer is responsible for about 10.4% of all cancer-related deaths in females ageing between 20 and 50 [1][3] . As observed in the figure 1, statistics in India show that it accounts for the majority of newly diagnosed cases of cancer and cancer-related fatalities. In the modern world, it must be taken into consideration as a crucial health issue. Specialist physicians have discovered various elements, such as lifestyle, hormonal, and environmental factors, that may raise a person's risk of getting BC. A DNA mutation that has impacted numerous generations of the family affects more than 5%–6% of BC patients. Additional reasons of BC include old age, obesity, and abnormal postmenopausal hormone levels. These problems have already been addressed by a number of ways, but the accuracy of those strategies has been constrained by noisy data. The dataset used in the proposed method is Wisconsin Diagnostic Breast Cancer (WDBC) that accurately classify BC. Because of this, the main objectives of this research are to use cancer disease features and a good preprocessing model to improve model accuracy for BC diagnosis and predict cancer affection at a preliminary phase. This study's main objective is to offer a simple way for detecting BC. This study meticulously examines current cancer detection techniques, producing amazingly precise and effective results. The structure for the following sections of the article is as follows. According to data published in December 2020 by International Agency for Research on Cancer (IARC), BC took over Lung Cancer as the leading and most common cancer among women worldwide. Overall number of cancer diagnoses has than doubled over the last two decades, rising from a projected 10 million in 20

FAQs

What problem does the paper address?

The paper addresses early breast cancer prediction to improve diagnosis and treatment by enabling more accurate identification in initial stages.

Which models are compared for breast cancer detection?

The study compares multiple machine learning models including SVM, KNN, Naïve Bayes, Logistic Regression, Random Forest, Decision Tree, XGB Classifier, and a deep learning ANN model.

How are features selected before training the models?

A feature-selection technique is used to choose crucial database properties, and correlation helps select the most correlated features from the data.

Aug 3, 2026

Upload by Olivia Brown