Efficient Method AND Algorithm for Preliminary Processing OF Medical Data

To solve medical problems that are practical, a special type of computer software is utilized. The ideal solution depends on the quality and volume of medical data. Nonetheless, most of the medical records are all based on sick leaves. Often, the collected medical data comprises the type information with a very lowered level of accuracy, non-standard terminology, information that is not incomplete and that is input incorrectly. Frequently, the patient's cardinal activity, behavior and habits are no how reflected in the medical data. Generally, there is no precise mechanism to collect medical data and dischargement of research in this sphere is one of the most necessitious issues ot consider. To solve these problems, algorithms that are based on full and partial selection as well as statistics have been used and these algorithms do not take intrinsic properties of the object into account while it is based on the interdependence of medical data. The algorithm which is proposed in the article eliminates these shortcomings and aims to reveal hidden connections togetherh with relationships between characters based on a random selection of the characters. This paper addresses topics (or issues) such as the initial processing of medical data generated by medical professionals, reclassifying learning choices/selections and choosing a range of informative characters (signs) that distinguish class objects. The underlined topics (or issues) are solved using algorithms to calculate grades based on Fisher criteria. As a result of primary data processing, five class objects are established (or formed): 1) Progressive angina pectoris; 2) Acute myocardial infarction (MI); 3) Arrhythmic form (AF); 4) Postinfarction cardiosclerosis (PC); 5) Permanent form of atrial fibrillation (PersAfib). Then problems such as the classification using Fisher's criteria and algorithms to calculate grades, the selection of a set (number) of informative characters (signs) that distinguish class objects, etc. are solved and appropriate software are developed. As key result, reference classes are established from the initial data processing. Objects that deviate from their class in the process of formation are excluded from the sample. Using established (or formed) classes, a set (or number) of classifications and indicators (i.e., informative characters) are selected. Initially there were five class objects of 62 characters each, all provided by medical professionals, and the initial processing of the data resulted in the formation of a reference table consisting of a set of 128 objects in first class, 110 in second class, 39 in third class, 76 in fourth class and 26 in fifth class, with 62 characters each.

Перевод пока недоступен