Assessing the Efficiency of Machine Learning Techniques on Multiple Cloud Platforms for Detecting Breast Cancer

Breast cancer is a serious global concern, accounting for 12.5% of all new cancer cases every year. Half of all breast cancer cases are diagnosed in women who have no recognized risk factors. The fundamental causes of breast cancer remain unknown, and no definitive cure has been discovered. Inadequate health-care resources frequently result in delayed diagnoses and treatments. This problem can be solved via machine learning. Advanced deep learning algorithms can provide early diagnoses, which can improve patient outcomes and boost survival rates dramatically. However, the main objective of this paper is to assess the computational power of the three different cloud platforms such as Google Cloud Platform (GCP), Amazon Web Services (AWS), and Microsoft AzureML (AML) over the range of algorithms namely Support Vector Machines (SVM), Naive Bayes (NB), Decision Trees (DT), Neural Network Classifier (NNC), and Logistic Regression (LR). The reference dataset for testing and comparison is the Wisconsin Breast Cancer Database (WBCD). Binary Neural Network has the highest accuracy and recall, with scores of 97.20% and 95.26%, respectively. Logistic Regression, on the other hand, displayed the highest precision, with a score of 99.13%. At 85.99%, the Boosted Decision Tree regressor had the greatest coefficient of determination for a regression model. Overall, across all the three cloud platforms chosen for the study Support Vector Machine has taken highest run time across the cloud platforms. The outcome shows that classification ML algorithms outperform regression ones for binary disease diagnosis like breast cancer, while regression models are less accurate. Also, the study suggests further research with a larger dataset and diverse ML algorithms for more conclusive insights into computational power on cloud platforms.