Application of Machine Learning Techniques for Diagnosis of Diabetes Based on Iridology

Abstract

Complementary and alternative medicine (CAM) is a system and therapy in the medical field that works based on knowledge, abilities, and practice. CAM is used to maintain health, diagnose disease, or to prevent and treat mental and physical illness. This technique can predict and treat disease. At the same time, machine learning has been widely used in the application of the biomedical field as a tool for diagnosing disease. The purpose of this work is to validate the use of iridology as a valid scientific technique to diagnose diabetes disease. Iridology combined with machine learning to simplify the diagnose process. Iris images were captured using Camera Iriscope Iris Analyzer Iridology. The region of interest (ROI) was cropped according to the location of the pancreas organ on iridology chart. The Gray Level Co-Occurrence Matrix method has been implemented for feature extraction. Five different classifiers method is used to classify diabetic and non-diabetic classes. The results are then validated and evaluated by using the k-fold cross-validation and confusion matrix, respectively. The subject consisted of two groups: one was 16 subjects non-diabetic and 11 subjects diabetic. The results show that the best accuracy is 85.6%, with specificity is 0.90, and the sensitivity is 0.80.