Analysis of computational techniques for diabetes diagnosis using the combination of iris-based features and physiological parameters
Digital image processing and advanced machine vision techniques are popular for the diagnosis of disease(s) in biomedical science. This paper presents a detailed comparative analysis of machine learning-based classification techniques to diagnose type 2 diabetes using the combination of iris-based features and physiological parameters. A set of 334 subjects are investigated which are divided into diabetic and non-diabetic groups. Moreover, the diabetic group is classified into three different subgroups according to the duration of the diabetic state. Statistical features, gray-level co-occurrence matrix, and gray-level run length matrix-based features are extracted from the specific areas of iris. Nine classifiers of different application areas are selected, and subsequently, six parameters (accuracy, precision, sensitivity, specificity, F-score, and area under the curve) of each classifier are analyzed. The analysis provided promising results with more than 95% of accuracy. The proposed technique can be used as a noninvasive and non-contact diabetes diagnosis tool which can help to find out the duration of diabetes in patients and the prevalence of diabetes.
Neural Computing and Applications