A Hybrid Deep Learning Approach for Thyroid Disease Diagnosis Using 1D CNN And Transformers

Thyroid disease is one of the most prevalent endocrine disorders, affecting millions of people worldwide and posing significant health challenges.Traditional diagnostic methods, such as clinical examinations and blood tests, are often time-consuming, reliant on practitioner expertise, and subject to variability.While existing machine learning (ML) and deep learning (DL) models have shown promise in improving diagnostic accuracy, they often struggle with poor generalizability, dataset variability, and the inability to simultaneously capture spatial and sequential features.This study proposes an innovative deep learning architecture that combines a 1D Convolutional Neural Network (1D CNN) with a Transformer model to enhance thyroid disease detection.The CNN efficiently captures spatial features, while the Transformer extracts long-range dependencies and contextual relationships, improving the model's overall understanding of complex patterns.The model was trained and evaluated on a thyroid disease patient dataset sourced from Kaggle, which includes demographic, medical, and laboratory data.It achieved an accuracy of 97.88%, surpassing existing models and highlighting its potential for early, accurate, and automated thyroid disease diagnosis.These findings demonstrate the model's capability to support better clinical decision-making and enable more effective personalized treatment strategies.