Information Technology for Medical Data Stream Mining

In this paper information technology for medical data stream analysis under conditions of uncertainty is proposed. In this case, the uncertainty is defined like unknown total number of patients, of the initial number of medical features and diagnoses that can be changed during diagnostic process, the data need to be processed sequentially in online-mode (Data Stream processing). Information technology consists of three modules: controlled learning module (Data Stream consists of sufficient number of marked samples - representative dataset), active learning module (Data Stream consists of a few number of marked samples - unrepresentative dataset) and self-learning module (Data Stream consists of only unmarked samples). As a result of the operation of information technology, we obtain a diagnosis of each patient in sequential online mode.