Infrared thermography (IRT) is a promising tool in the medical field, particularly for the non-invasive detection of thermal anomalies associated with diabetic foot complications, such as ulcers, peripheral neuropathy, and vascular disorders. Although IRT has shown increasing effectiveness in identifying temperature variations linked to these conditions, the integration of artificial intelligence (AI) for the classification of at-risk groups is still underutilized. In this study, we use the VGG16 model, a convolutional neural network (CNN) recognized for its performance in the medical domain, to classify diabetic patients from thermal images. The input of the classifier is the thermal image and the output is the ischemic and or not-ischemic group. The VGG16 model achieved an accuracy 88.29%. This result emphasizes the importance of AI, combined with IRT, to improve early diagnosis of diabetic foot and strengthen preventive care.