Abstract – The increasing occurrence of Type II Diabetes Mellitus (T2DM) due to lifestyle changes has required the development
of non-invasive diagnostic methods. This study explores the potential of iridology, another diagnostic approach, joined with
machine learning (ML) algorithms to detect T2DM based on a precise Region of Interest (ROI) in the right iris. This work
introduces two ML-based classification methods. The first method employs multiple ML models with changing K-fold cross-
validation (ranging from 2 to 20 folds), achieving a maximum classification accuracy of 78.5% at 5-fold using Support Vector
Machine (SVM) and Binary Generalized Linear Model (GLM) Logistic Regression. The second method employs Principal
Component Analysis (PCA) to enhance feature selection, improving accuracy to 82.2% by training predictions from two initial
classifiers-Coarse Decision Tree (74.8% at 14-fold, PCA variance 97%) and Linear Discriminant Analysis (77.6% at 9-fold,
PCA variance 100%)-before refining classification with a Binary GLM Logistic Regression model. The proposed approach offers
a promising, non-invasive alternative for early diabetes detection using iris analysis and Artificial Intelligence (AI)
Full Abstract: https://iridology-research.com/pdf/Non-Invasive_Diagnosis_of_Type_II_Diabetes_Using_I.pdf
Authors: Alaa Abdulkareem Ahmed, Mohammad Tariq Yaseen
Department of Electrical Engineering, College of Engineering, University of Mosul, Iraq.
