Hemoglobin Estimation and Anemia Severity Prediction Using Machine Learning Algorithms

Anemia is one of the most prevalent and widespread health issues worldwide, according to the World Health Organization (WHO). It occurs as a result of hemoglobin or red blood cell deficiency. The National Cancer Institute categorizes anemia severity as mild, moderate, severe, and life-threatening. Predicting the severity of anemia is vital for effective treatment decisions, how it affects health, and monitoring progress over time. In this study, we estimated the level of hemoglobin and predicted anemia severity using various machine learning (ML) methods. To estimate the level of hemoglobin and predict the severity of anemia, we developed regression- and classification-based ML models using the Complete Blood Count (CBC) values collected from Bangabandhu Sheikh Mujib Medical University (BSMMU). Before training the models, the data were reduced, balanced, and standardized. The outputs of various regression and classification models were compared with each other. In this research, Neural Network showed better performance with high accuracy and less error for the estimation of hemoglobin level compared to other regression models. When predicting the severity of anemia, Random Forest outperformed other classification models with high accuracy.