Hybrid Model for Detecting Anemia from Palm Images using Vision Transformers and Machine Learning

Taipe M.; Ascurra K.; Ticona W. · Lecture Notes in Networks and Systems (2026) · DOI: 10.1007/978-3-032-20746-3_13

Resumen

Anemia is a critical public health problem in developing countries, where conventional diagnostic methods are invasive and poorly accessible. This work presents a hybrid model for non-invasive anemia detection from palm images, using the public dataset “Anemia Detection Using Palpable Palm Image” from Ghana. Deep learning-based feature extractors (BoTNet, ViT, Swin Transformer, PiT, and MobileViT) were evaluated in combination with machine learning classifiers (SVM, Random Forest, k-NN, Naïve Bayes, and Decision Tree). The best performance was obtained with MobileViT + SVM, achieving 99.75% accuracy, 100% precision, 99.5% recall, an F1-score of 99.75%, and an AUC of 1.0. These results demonstrate the potential of lightweight hybrid models as an accurate and scalable alternative for early anemia detection in resource-limited settings. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Instituciones: Universidad ESAN; Universidad Tecnológica del Perú