Smart Municipal Services and Predictive Healthcare: A Thermal Imaging Perspective

Abstract

In the present day, diabetes is viewed as a serious problem. It can be promoted and campaigned as a smart municipal service to create awareness. The heart, nerves, eyes, and other human disorders, among others, might all be negatively impacted by this illness. Thus, early detection of diabetes patients is crucial to implement preventative treatments as soon as possible. In this work, a machine intelligence (MI) based approach is proposed for classifying diabetic and non-diabetic patients from the thermal image analysis of the human foot. This approach is focused on several machine learning (ML) models, such as k-nearest neighbour (KNN), decision tree (DT), AdaBoost (AB), and Naive Bayes (NB), to carry out such classification mechanisms. This study employs a cross-validation mechanism with the number of folds (NFD) set to 3, 5, and 10. By analysing the percentage of classification accuracy (CA) based on the dataset for different ML-based models, KNN achieved superior classification results than DT, AB, and NB, which are 93.30%, 94.60%, and 95.10%, for NFDs 3, 5, and 10, respectively.