An Advanced Diagnostic Framework for Thyroid Disease Prediction using Enhanced Deep Neural Networks

Machine learning models have recently become valuable tools for diagnosing thyroid conditions. It offers potential improvements in diagnostic accuracy; this study examines the effectiveness of Artificial Neural Networks for thyroid disease classification and imposes several dimensionality reduction and feature selection techniques, which include Linear Discriminant Analysis, Fisher kernel Discriminant, and Principal component analysis. Random Forest assesses feature importance and establishes an input feature set to clarify model performance further. The result indicates that integrated ANN with these preprocessing techniques significantly increases classification accuracy. This suggests a methodology for the clinical diagnosis of thyroid diseases. The research underscores the effectiveness of ANN and Data Preprocessing in improving diagnostic precision for thyroid related conditions.