Machine learning techniques to classify blood groups and Rh-factors

• Main Author: Mahmood, Mustafa F.
• Format: Conference Proceeding
• Language: English
• Subjects: Accuracy; Algorithms; Blood groups; Blood transfusion; Body fluids; Classification; Error analysis; Error detection; Erythrocytes; Human error; Machine learning; Neural networks; Support vector machines; Testing time
• ISSN: 0094-243X; 1551-7616
• DOI: 10.1063/5.0236272

Blood is an important part of body fluids. The red blood cell is one of the parts of blood. Blood transfusions require the provider to know the recipient’s blood type. The current method for determining blood type depends on a medical laboratory technician. This method is subject to errors in writing or reading the result, and thus, the error can be fatal. This study aims to create and test a blood group detection and classification prototype that can eliminate the possibility of human mistakes. The suggested method employs machine learning techniques to detect and classify blood groups. The system consists of practical and theoretical units. The practical unit includes collecting samples using main tools. The theoretical unit includes the process of using machine learning using four techniques: Logistic Regression (LR), Neural Network (NN), K-Nearest Neighbors Algorithm (KNN), and Support Vector Machines (SVM) in the orange program. The experimental results were conducted on 1300 raw data and using 50 samples using techniques LR, NN, KNN, and SVM. The experimental results proved that the accuracy for classifying the samples was 100% using techniques LR and NN, and the accuracy for classifying the raw data was 98% for each. In addition, the test time using Technique LR was 1.3 seconds. This study achieved high accuracy compared to previous studies according to the outcomes.