Informatica

The main purpose of this article was to compare traditional binary logistic regression analysis with decision tree analysis for the evaluation of the risk of cardiovascular diseases in adult men living in the city. Patients and methods. In our study, we used data from the Multifactorial Ischemic Heart Disease Prevention Study (MIHDPS). In the MIHDPS study, a random sample of male inhabitants of Kaunas city (Lithuania) aged 40–59 years was examined between 1977 and 1980. We analysed a sample of 5626 men. Taking blood pressure lowering medicine, disability, intermittent claudication, regular smoking, a higher value of the body mass index, systolic blood pressure, age, total serum cholesterol, and walking in winter were associated with a higher probability of ischemic heart disease or cardiovascular diseases. Having more siblings and drinking alcohol were associated with a lower probability of these diseases. The binary logistic regression method showed a very slightly lower level of errors than the decision tree did (the difference between the two methods was 2.04% for ischemic heart disease (IHD) and 2.86% for cardiovascular disease (CVD), but for consumers, the decision tree is easier to understand and interpret the results. Both of these methods are appropriate to analyse cardiovascular disease data.

Data stream mining has become a novel research topic of growing interest in knowledge discovery. Most proposed algorithms for data stream mining assume that each data block is basically a random sample from a stationary distribution, but many databases available violate this assumption. That is, the class of an instance may change over time, known as concept drift. In this paper, we propose a Sensitive Concept Drift Probing Decision Tree algorithm (SCRIPT), which is based on the statistical X² test, to handle the concept drift problem on data streams. Compared with the proposed methods, the advantages of SCRIPT include: a) it can avoid unnecessary system cost for stable data streams; b) it can immediately and efficiently corrects original classifier while data streams are instable; c) it is more suitable to the applications in which a sensitive detection of concept drift is required.

Application of knowledge discovery in databases (data mining) for medical decision support is discussed in this work. The aim of the study was to use decision support algorithm for the differential diagnosis of intraocular tumors using parameters from eye images obtained by the ultrasound examination. Application of predictive modeling algorithm for decision tree formation using See5.0/C5.0 data mining system is presented. The decision tree was build using tumor geometry and microstructure parameters. The use of decision tree allows to differentiate tumors from other tumor-like formations. Low percentage of diagnostic errors shows that decision tree is reliable enough to offer it as “second opinion” for physician's decision support.