A new method of data preparation for cardiological decision support

There has been huge progress in the introduction of new digital methods, such as decision support, in cardiology. Data preparation is the most important and the most time-consuming part of the data mining process. We present a newly developed hierarchical method of text classification based on regular expressions. This method is the basis of our data mining system during the preprocessing stage to transform Latin-based free-text medical reports into a decision table. In this study we also compare the accuracy and scalability of our method with an approach based on dictionary phrases.