In the human body, the most important organ that regulates the metabolism of the human body is the thyroid gland. This gland discrete the specific hormones through the body control itself and works properly. There two types of thyroid disorders are hypothyroidism and hyperthyroidism. If a person is affected by both of these disorders, the thyroid gland discrete a special type of hormone into the human body's blood so that the regulatory system could be disturbed by this. Different analytics tools are used for developing a risk of thyroid disease, but firstly cleaning the data. In this current study for the implementation and detecting the disease dataset has been taken from the UCI Fuzzy neural network Repository. The reason of formulating the unnecessary or less hormones is occurred due to the iodine deficiency or other inflammatory conditions. Data samples are enough to evaluate the results. Fuzzy neural network plays a major role for disease detection. This paper employed various MLand Fuzzy neural network methods and techniques for the diagnosis and preventing the disease. In this proposed work walrus grey algorithm is used to choose the discriminative features of thyroid disease. Feature selection is conducted with the hybrid method using the walrus algorithm optimization. The principal component of the encoders used in this methodology is the patch-based encoder. The basic disease mechanism is diagnosed and detected in thyroid by its symptoms. The WAORF guided the features that identify the optimal solutions. Significant improvements across the performance metrics are used to compute the results for achieving better results.