Statistical Modeling on Visual Detection of Anemia

Anemia, a prevalent blood disorder, disproportionately impacts vulnerable populations globally. Traditionally, its diagnosis relies on invasive blood tests that measure hemoglobin levels, necessitating laboratory equipment and skilled personnel. This requirement often restricts access in resource-limited settings. This has spurred interest in non-invasive diagnostic methods, especially those using imaging technologies. This review focuses on image-based methods and the role of statistical modeling. Visual signals from images, such as pixel intensity values, are transformed into predictors and used in models like logistic regression (LR), support vector machines (SVM), decision trees (DT), and k-nearest neighbors (kNN). Among them, kNN is often the most precise and interpretable, especially in studies with small sample sizes. Still, challenges remain. Variations in image quality, lighting, and preprocessing can affect model performance, and visual features may not detect all anemia types. Further work is needed to improve standardization and expand validation. Overall, statistical models help transform image data into meaningful clinical insight and support the development of accessible, non-invasive anemia diagnostics.