生物医学データマイニングの国際ジャーナル
オープンアクセス

概要

Different Mining Techniques for Health Care Data Case Study of Urine Analysis Test

Mohamed D Almadhoun* and Alaa M El-Halees

To make huge amounts of data that is produced by health care information systems useful and important to the potential, we apply knowledge discovery. This study considers urine analysis test results as an input data to different data mining techniques in order to discover the hidden and meaningful patterns in data. It also shows results of evaluation and analysis for data patterns. Data mining techniques were, 1) Classification to support functionality for alerting about new instance that does not match the predicted value by classification model, 2) Association rules to inform about relations and changeability between elements, 3) Clustering to categorize patients into separate groups to give an indication about how to deal with each patient, and 4) Outlier analysis to discover the most sick patients or unfamiliar cases that need a special care. Resulting knowledge were novel, actionable, understandable, and valid. This was stated by applying two methods of evaluation, first was a survey about results and was filled by medical specialists, second was by cross validation and T-Test.