|
Garipov Ildar Ildarovich Faculty: Computer information technologies and automatics (FCITA)
|
Abstract
The concept of the automated workplace of medical staff has arised after appearance of computers and medical equipment with computing functions. This work is about creation of specialized computer system of fat embolism syndrome diagnostics.
Fat embolism is plural occlusion of blood vessels with drops of fat. This is one of the most terrible and heavy complications of the early period of traumatic illness.
Hospital lethality among victims with a mechanical injury remains high, making from 25 up to 59 % according to different authors. The reason for this is both severity of received injuries, and development of complications which become very important during traumatic illness and define it outcome.
Nowadays the mechanisms of fat embolism development are not fully investigated and this leads to high lethality. In some cases this involve to inefficient preventive measures and treatment.
In modern literature we can see narrow questions concerning diagnostics and treatment of fat embolism syndrome (FES). The suggested trouble-shooting tests are directed for discovering clinical symptoms of FES. Only some works present the problem of a fat embolism syndrome conceptually, from positions of the doctrine about traumatic illness and looking at the metabolic faults occurring in an organism after getting injury.
The urgency of such investigation in medical diagnostics is obvious. The requirement for the qualified experts in many medical specialities is much more than the proposals.
Non-obvious problems of medicine and biology were ideal ground for using neural network technologies and in this area we can see the best practical success of neural informational methods.
In supplement to medical diagnostics neural networks enable to increase specificity of method without reducing its sensitivity. Distinctive property of neural network is that they are not programmed, do not use any conclusion rules for definition of diagnosis and are trained to do this on examples. In the most of diagnostics tasks, differential diagnostics, prognosis, choice the strategy and tactics of treatment, etc. It is very easy to get necessary amount of examples for training neural networks. Medical tasks always have several ways of the decision and answer which is not define clear and coincides with the way of neural network result delivery.
Diagnostics is a special case of events classification. The greatest value is presented with classification of those events which are absent in the set which train neural network. Here is the advantage of neural network. It is possible to do such classification, generalizing previous experience and using it in new cases with help of neural network.
The software which is created with specialized computer system will contain digital X-ray processing and complex analysis of laboratory, clinical and medical history parameters used in diagnostics of FES. Using technologies of knowledge processing allow to bring line of the basic advantages to diagnostic system which are absent in usual prototypes. First, it is an opportunity of using and accumulation high-quality knowledge and experience of the most qualified experts, experts of a concrete subject area (SA). Second, an opportunity to adapt system according to knowledge changes about SA. Third, an opportunity of intellectualization interaction with user according to dispose of his concrete features.
In scientific literature is described the set of various parameters of homeostasis condition which were applied for diagnostics and prognosis of fat embolism. That makes the problem difficult and confusing.
Variety of parameters and their estimations compel to systematize changes which were found and to pick out the most authentic attributes, resulting various compositions of diagnostic parameters.
Approaches for diagnostics FE can be grouped as follows:
• Transfer of the basic attributes which should be the most important for diagnosis determination;
• Transfer of attributes with reduction of occurrence frequency;
• Division of attributes on main and minor;
• Combinations of the main and minor attributes with a conclusion to the formula of sufficiency for the diagnosis.
Employees of a scientific research institute of traumatology and orthopedy of Donetsk state medical university named by M.Gorkiy have been developed a technique for diagnostics of this disease. Using it on practice help to diagnose and to reveal correctly this disease. Correctness in diagnosis determination was 90 % what is is a good parameter for the medicine. The neural network of return mistake spreading. It reacts to entrance influences and provides determination of the diagnosis of fat embolism syndrome. That proves correctness of the chosen method. Software will hold experience and knowledge of leading experts in this area. So using neural network in diagnostics of FES and processing of digital X-rays processing will allow to create effective SCS. Program which is doing during master’s scientific work is unique as it has no analogues nowadays.
|