Comparative Evaluation of Classical Machine Learning Algorithms in Detection of Beta Thalassemia Minor Carrier

Abstract

Thalassemia, a significant public health issue in Mediterranean countries including Turkey, is one of the preventable hereditary disorders. It is a genetic disease that arises due to a defect in the synthesis of hemoglobin—the protein responsible for carrying oxygen molecules in the blood to tissues and organs—resulting in either reduced or absent production of polypeptide chains. Thalassemia has four clinical forms: Thalassemia Major, Thalassemia Intermedia, Thalassemia Minor, and Silent Carrier. Individuals with Thalassemia Major and Intermedia are considered patients and often face various health challenges throughout their lives. In contrast, Silent Carriers and those with Thalassemia Minor are merely carriers and typically do not experience any health issues related to the condition. The disease primarily manifests in children born to two carrier parents. In this study, machine learning models were developed using complete blood count data obtained from primary healthcare centers within the scope of a premarital screening program. The aim was to identify individuals who are not suspected of being thalassemia carriers without the need for HPLC testing. The performance of the proposed models was evaluated and compared using key metrics such as accuracy, precision, recall, and F1-score.

Ähnliche Arbeiten