Anemia detection through non-invasive analysis of lip mucosa images.

This paper aims to detect anemia using images of the lip mucosa, where the skin tissue is thin, and to confirm the feasibility of detecting anemia noninvasively and in the home environment using machine learning (ML). Data were collected from 138 patients, including 100 women and 38 men. Six ML algorithms: artificial neural network (ANN), decision tree (DT), k-nearest neighbors (KNN), logistic regression (LR), naive bayes (NB), and support vector machine (SVM) which are widely used in medical applications, were used to classify the collected data. Two different data types were obtained from participants' images (RGB red color values and HSV saturation values) as features, with age, sex, and hemoglobin levels utilized to perform classification. The ML algorithm was used to analyze and classify images of the lip mucosa quickly and accurately, potentially increasing the efficiency of anemia screening programs. The accuracy, precision, recall, and F-measure were evaluated to assess how well ML models performed in predicting anemia. The results showed that NB reported the highest accuracy (96%) among the other ML models used. DT, KNN and ANN reported an accuracies of (93%), while LR and SVM had an accuracy of (79%) and (75%) receptively. This research suggests that employing ML approaches to identify anemia will help classify the diagnosis, which will then help to create efficient preventive measures. Compared to blood tests, this noninvasive procedure is more practical and accessible to patients. Furthermore, ML algorithms may be created and trained to assess lip mucosa photos at a minimal cost, making it an affordable screening method in regions with a shortage of healthcare resources.