Technique to Identify & Classify Thyroid Cancer Using Supervise & Supervise Learning

Main Article Content

Megha Jain, Mr. Nirmal Singh, Dr. Vikas Somani, Dr. Awanit Kumar


Background: Thyroid cancer is caused by thyroid gland, located in front part of neck below voice box called larynx. The thyroid gland is an endocrine gland, which controls hormones in the human body resulting the regulation of metabolism. It is one of the common cancer, which is about four per thousand people in the world. Thyroid cancer is seen more in women as compare to men as it is 1.9 and 6.1 per hundred thousand people in the world. From past ten years in India, it has been seen that the growth rate is increased up to 62 percent and 48 percent in women and men respectively.

Objectives: To develop and compare machine learning algorithm for thyroid cancer detection and prevention at early stage.

Methods: Artificial intelligence (AI) is continuously changing the shape of the healthcare industry. To categorize the cancer types in thyroid disease, a machine learning-based support vector machine approach has been applied. The multiclass support vector machine is one of the best-performing approaches in this domain. The support vector machine algorithm has been applied to the machine learning datasets offered by the university of California Irvine to evaluate (train, test, validate) the final AI model.

Results: The final AI model using the support vector machine technique shows the precision is 73 percent, recall is 81percent , and f1-score is 77 percent respectively. The result is very significant in this domain with overall 97 percent accuracy.

Conclusions: Thyroid cancer is among the most frequent types of cancer. It is really difficult to detect it at an early stage. Machine learning (ML) is an emerging method for thyroid cancer categorization and prediction. In this paper machine learning-based multiclass support vector machine approach has been used to categorize the types of thyroid cancer with a 97 percent accuracy.

Article Details