Prediction of foot risk classification for Type II Diabetic through image analysis

The global prevalence of diabetes mellitus has increased. The use of electronic platform devices has grown in popularity due to their low cost and ease of use. Despite their benefits, however, there remain concerns about their accuracy and precision. The objective of the study is to determine the accuracy and precision of the Win-track platform. In a cross-sectional study, 49 male patients’ data were collected. Based on the pressure asserted, the data were further classified into different stages from 1 to 4. The study used four different types of classifiers (Logistic, Multi-layer Perceptron, Simple Logistic Regression and Meta-logit Boost) to check the accuracy. The result shown for all the classifiers was positive with Meta-logit Boost giving the higher Mathews correlation coefficient (MCC) (stage 1=1, stage 2=1, stage 3=0.904 and stage 4=0.912) and highest correctly classified instances and lowest incorrectly classified instances (95.91% and 4.08% respectively) with least amount of time taken for execution (T = 0.02ms). With respect to the accuracy obtained, it is suggested to use the Win-track platform in hospitals and clinics.