Early Prediction of Thyroid Cancer using Hybrid Combination of Swarm Optimization and Meta Classifier based Machine Learning Algorithm

Thyroid cancer is one of the most common endocrine cancers worldwide, demanding reliable prediction models for early identification and treatment. In this work a new technique to thyroid cancer prediction has been proposed that combines particle swarm optimization (PSO), genetic algorithms (GA), and a meta-classifier. The goal is to use the optimization capabilities of PSO and GA to extract an ideal subset of features and tweak the hyper parameters of base classifiers and the meta-classifier, hence improving prediction performance. The suggested technique intends to offer doctors with a reliable tool for predicting thyroid cancer, allowing for early identification and intervention. The model optimizes the information in the data by integrating PSO, GA, and a meta-classifier, hence enhancing projected accuracy and clinical value. Common measurements used to evaluate the model's performance include accuracy, precision, recall, F1-score, and ROC-AUC. The generalizability and robustness of the proposed model is further improved through Cross-validation technique. The experimental findings show that the proposed strategy outperforms previous approaches with 98.65% accuracy.