A Hybrid Model for Thyroid Disease Classification Using Evolutionary Multivariate Kernal SVM Prediction Method

Thyroid diseases are widespread worldwide. In India too, there is a significant problems caused due to thyroid diseases. Various research studies estimates that about 42 million people in India suffer from thyroid diseases. There are a number of possible thyroid diseases and disorders, including thyroiditis and thyroid cancer. This paper focuses on the classification of two of the most common thyroid disorders are hyperthyroidism and hypothyroidism among the public. The National Institutes of Health (NIH) states that about 1% of Americans suffer from Hyperthyroidism and about 5% suffer from Hypothyroidism. From the global perspective also the classification of thyroid plays a significant role. The conditions for the diagnosis of the disease are closely linked; they have several important differences that affect diagnosis and treatment. The data for this research work is collected from the UCI repository which undergoes pre-processing. The pre-processed data is multivariate in nature. Curse of Dimensionality is followed so that the available 21 attributes is optimized to 10 attributes using Hybrid Differential Evolution Kernel Based SVM algorithm. The subset of data is now supplied to Support Vector Machine (SVM) classifier algorithm where Radial Basis Function Kernal (RBF) is used. In order to stabilize the errors this iterative process takes 30 runs and the data is classified. The accuracy of classification is observed to be 67.97%. This result is average when compared to our previous work that used the Kernel based Naïve bayes classifier.