Case-based reasoning (cbr)-based anemia severity detection system (asds) using machine learning algorithm

Abstract

Anemia is a very common disease around the world. Based on the patient’s signs, symptoms and hemoglobin amount of blood, the severity of anemia can be detected. This paper proposes an Anemia Severity Detection System (ASDS). The ASDS is designed to help the doctors or medical experts to detect the severity of anemia in a patient. It is developed based on Case-Based Reasoning (CBR) methodology and machine learning algorithm. CBR is one of the popular artificial intelligence techniques that uses past experiences to derive results for new cases, and it works in a cycle that includes four activities: Retrieve, Reuse, Revise and Retain. As a machine learning algorithm, K-Nearest Neighbor algorithm is used to detect whether the anemia is mild or severe. The dataset which helps to develop ASDS is from the Website of IUScholarWorks. As a screening tool, the graphical user interface (GUI) of ASDS has been developed with the help of Python, so that doctors or medical experts can access the system through the GUI. To improve accuracy and reduce training time, data preprocessing technique and feature selection technique have been applied on the dataset. Experimental results show that after using both techniques, K-Nearest Neighbor model gives highest 92% accuracy, 84% precision and 90% recall results.