Fedorenko M. Neural network models and structures of a multilevel information-analytical system for diagnosing urological diseases

Українська версія

Thesis for the degree of Candidate of Sciences (CSc)

State registration number

0421U101789

Applicant for

Specialization

  • 05.13.06 - Інформаційні технології

30-04-2021

Specialized Academic Board

Д 64.062.01

National Aerospace University "Kharkiv Aviation Institute"

Essay

Object of research - the processes of data collection and processing in the diagnosis of diseases in urology; the purpose of the research is to develop neural network models and create on their basis a multilevel information-analytical system for reliable diagnosis of urological diseases; research methods - mathematical apparatus of set theory and probability theory, methods of neural network modeling, methods of mapping theory and mathematical statistics, methods of systems analysis and theory of algorithms, methods of reliability theory; results - methods, models and structures of information technology for diagnosing diseases in urology, which allow by increasing the parameters of uroflowmetry and ensuring the exchange of reliable diagnostic information between specialized medical institutions to teach neural network modules to increase the reliability of diagnosis; novelty - for the first time a multilevel structure of neural network modules of information-analytical system for diagnosing urological diseases is proposed, the training of which, unlike known ones, is based on operations of exchange and modification of data on diagnoses of local, regional and national levels use of additional information for additional training; improved neural network model of disease recognition in urology by taking into account uroflowmetric parameters of patients, in particular: type of uroflowgram, maximum and average flow rate, as well as the location of their values on nomograms, which reduces the risk of misdiagnosis; further developed the method of redundancy of information-analytical system for diagnosing urological diseases by exchanging diagnostic data by using structural redundancy of means of realization of neural network modules at different levels of the system hierarchy, which allows to increase its reliability; degree of implementation но introduced in Kharkiv Medical Academy of Postgraduate Education, National Aerospace University. "Kharkiv Aviation Institute"; industry - diagnosis of diseases in urology.

Files

Similar theses