Machine Learning-Assisted Cervical Cancer Prediction Using Particle Swarm Optimization for Improved Feature Selection and Prediction

Abstract

Cervical cancer is a common and deadly disease that affects women worldwide. Early diagnosis and treatment can improve the survival and quality of life of patients. Machine learning techniques can help to analyze complex and high-dimensional data related to cervical cancer and provide accurate and reliable predictions. However, selecting the most relevant and informative features from the data is a challenging task that affects the performance and interpretability of machine learning models. This paper proposes a novel method that uses particle swarm optimization (PSO) to perform feature selection and optimization for cervical cancer prediction. PSO is a bio-inspired algorithm that mimics the social behavior of a swarm of particles that search for the optimal solution in the feature space. The use of PSO to select the best subset of features that maximize the classification accuracy of eight machine learning models: Support Vector Machines (SVM), Gaussian Naive Bayes (GNB), Random Forests (RF), Decision Trees (DT), Extreme Gradient Boosting (XGB), Linear Regression (LR), Adaptive Boosting (AdaBoost), and K-nearest neighbor (KNN).To evaluate the method, a publicly available dataset was used, the Cervical Cancer Risk Factors Dataset (CCRFD). Then, compare the results with several state-of-the-art methods that use different feature selection techniques and ML algorithms. The experimental results show that the method achieves superior performance in terms of feature reduction rate, accuracy, precision, and AUC. Specifically, the Adaboost-PSO model performed best in terms of feature reduction rate with a reduction of rate 100% while the RF-PSO model performed best in terms of accuracy and precision, with an accuracy of 98% and precision of 100%.