Constructing a Data Mining Sensor to Classify Tw-DRGs Medical Specialties

Abstract

This work applies decision tree analysis to construct a classification sensor for Taiwan’s diagnosis related groups (Tw-DRGs) of medical specialties. Inpatient expenses of 2009 in a national health insurance database were transformed to the SQL database analytic software for analysis and classification. DRG code is selected by the principal diagnosis code with the operating room procedures code. Selecting the combinations in accordance with National Health Insurance inpatient diagnostic related groups TW-DRG Classification Manual, resulting in repeated DRG codes. The study analyzes three medical classifications of Tw-DRGs implementations on 2009 to solve the DRG code appropriate attribution issues. Appropriateness and accuracy of the DRG code is an important influence to the medical payments. The results found that decision tree can effectively and quickly establish the DRG classification process.