An investigation on machine learning models for enhanced thyroid prediction

The paper aims to enhance the prediction of thyroid diseases by optimizing the deep learning process and tuning hyperparameters. This project utilizes a dataset focused on one of the most critical issues in health care: thyroid disease diagnosis. Data preprocessing, including Z-scale normalization, was applied to reduce overfitting and ensure the significance of feature contributions. Hyperparameter tuning of machine learning (ML) techniques is primarily utilized for Recurrent Neural Networks (RNNs) to optimize training and improve model classification performance. Comparison variables are utilized in ML methods, specifically with the random forest technique, to enhance model performance and accuracy. This work showcases improvements in the analytical framework and establishes a foundation for more efficient and accurate detection of thyroid diseases.