Feature based analysis of endometriosis using machine learning

Abstract

Machine learning is a cutting-edge technology used for predicting and diagnosing various diseases. Various machine learning algorithm facilitates the prediction. The decision tree belongs to learning algorithm that performs both classification and prediction. The decision tree constructs the tree-like to evaluate the best features. The decision tree performs well in the prediction of various diseases. Endometriosis is a recurrence disease that creates an emotional impact in women. Endometriosis is a lump-like structure that appears at several locations in reproductive organs of women. The diagnosis of endometriosis was predicted through scanning procedures and laparoscopic procedures. The symptoms identified from laparoscopic surgery were used as the features for predicting the severity of endometriosis. The symptoms include mass-like structure, tissue size, variation in tissue colour, and blockages in fallopian tubes. The decision tree analyze the features of endometriosis by using two criteria such as entropy and Gini index. The entropy and Gini index construct the tree by identifying the size of tissue as major influencing attributes. The Gini index outperforms well with training accuracy of 84.08% and test accuracy of 84.85.