Anemia Prediction Using Machine Learning Approach

Anemia is a common illness with a major influence on world health, so early detection and treatment for anemia is very much necessary and efficient diagnostic methods are very much required. In addition to other pertinent clinical parameters, analysis of Red Blood Cell (RBC) prints can play a major role in early identification of anemia. RBC prints are analyzed using ML models in this paper, which offers a unique method for anemia prediction using hemoglobin count, percentage of Red, Green and Blue pixels of RBC prints and gender to produce an all-encompassing predictive model. The performance of ML models such as KNN, SVM etc. have been evaluated using cross validation technique by calculating the mean accuracy, f1, precision and recall scores for different folds. The best performing model evaluated through fifteen-fold cross validation is Decision Tree Classifier showing an accuracy of 98.72%.