A Search for Machinery Intelligence Towards Sustainable Health : An Improved Ensemble Cervical Cancer Diagnosis

Abstract

Health professionals identify cervical cancer as a potentially fatal condition. Patients' valuable lives are at risk due to the difficult late diagnosis and treatment. The formal screening for illness identification suffers in both developed and developing countries because of high medical costs, a lack of healthcare facilities, social norms, and the late onset of symptoms. Early detection of various different illnesses, including cervical cancer, is possible because to machine intelligence. It is also cost-effective and computationally cheap. Modern, time-consuming medical treatments are not necessary for the patients, and machine intelligence can help with early cervical cancer detection. The reliance on a single classifier's prediction accuracy is the issue with the present machine classification approaches for illness detection. Due to bias, over-fitting, improper treatment of noisy data, and outliers, the use of a single classification method does not guarantee the best prediction. In order to provide an appropriate diagnosis that addresses the patient's symptoms or problems, this research study presents an ensemble classification approach based on majority voting. A broad variety of classifiers, including Decision Tree (DT), Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbor (KNN), Naive Bayes (NB), Multiple Perceptron (MP), and Logistic Regression (LR) classifiers, are experimented in the study. The study shows an increase in prediction accuracy of 94%, which is much higher than the prediction accuracies of individual classification methods tested on the same benchmarked datasets. As a result, the suggested paradigm offers health professionals access to a second opinion for early illness detection and treatment.