Using Data Mining Algorithms for Thalassemia Risk Prediction

Abstract

This study predict the risk of thalassemia in all age groups based on identified risk of thalassemia. Knowledge about the risk factors for thalassemia was identified using structural interview with experienced medical personnel and questionnaire which was used to collect empirical medical database on the parameters. Supervised machine learning algorithms was used to formulate the predictive model for risk of thalassemia using the parameters and data identified and collected. The predictive model for the risk of thalassemia was simulated using the Waikato Environment for Knowledge Analysis (WEKA). The simulated model was validated using the historical data collected from the hospitals explaining the parameters and the risk of Thalassemia. The results of the study showed that following the collection of data from 51 patients, the parameters identified included demographic variables like gender, age, marital status, ethnicity and social class while the clinical variables included family history, spleen enlargement, diabetes, urine colour changes and parent carriers while the distribution of the risk was 43% no cases, 10% low cases, 16% moderate cases and 31% high cases. The study concluded that using the multi-layer perceptron for the prediction of Thalassemia will improve the decision making process within the healthcare service concerning Thalassemia.