Леонид ФайнзильбергНТУУ "КПІ імені Ігоря Сікорського", МННЦ ИТіС · Кафедра Біомедичної кібернетики
Леонид Файнзильберг
Doctor of Engineering
METHODS AND SYSTEMS FOR ARTIFICIAL INTELLIGENCE IN MEDICINE
About
39
Publications
429
Reads
How we measure 'reads'
A 'read' is counted each time someone views a publication summary (such as the title, abstract, and list of authors), clicks on a figure, or views or downloads the full-text. Learn more
79
Citations
Introduction
COMPUTER’S TOOLS OF PREVENTIVE MEDICINE FOR HOME USING
Publications
Publications (39)
The author formulates definitions of computer algorithm efficiency according to a criterion that characterizes accuracy, reliability, performance speed, and other consumer properties. Schemes of proof experiments based on stochastic models of generating artificial data with statistical characteristics adequate to real observations are suggested. Th...
The author formulates definitions of computer algorithm efficiency according to a criterion that characterizes accuracy, reliability, speed, or other consumer properties. Schemes of proof experiments based on stochastic models of generating artificial data with statistical characteristics adequate to real observations are suggested. The experiments...
The definition of the usefulness of the binary classifier from the point of view of reducing the a priori risk of false classification is formulated. Sufficient conditions are proposed to guarantee the utility of a diagnostic test according to this definition. The obtained conditions improved the traditional ROC analysis by limiting the correspondi...
The author develops an approach to constructing biometric systems based on the analysis of the phase portrait of a single channel electrocardiogram (ECG) of the test subject. The rules for solving the problem of human identification and verification (authentication) are proposed. Experimental studies have shown that the proposed decision rules prov...
Research in engineering, biology, economics and other areas is often associatedwith the analysis of observed processes that are repetitive in time. One of thepromising approaches to solving the problem of analyzing and interpreting suchsignals is based on converting the original cyclic signal into a sequence of symbols of some alphabet, for which m...
For early detection and timely correction of imbalances in the body that can lead to the development of various diseases, personalized devices are needed with which one can control the current state of the body at home. The purpose of the article is to develop a universal approach to the construction of such tools and, using examples of solving urg...
Introduction. An important area of modern information technology application is medical diagnostics, which is based on computer processing of biomedical signals The purpose of the article is to provide information on the results of basic and applied research that has ensured the practical implementation of the ECG method (fasegraphy method) in vari...
Introduction. The linguistic approach, based on the transition from the observed cyclic signal to a sequence of symbols (codeword), which characterize the dynamics of indicators from cycle to cycle, makes it possible to use the procedures of mathematical linguistics to increase the reliability of decisions. The purpose of the article is to expand t...
New methods for retrieving localized diagnostic information from cyclic signals of complex shape are proposed. The advantages of an alternative method of estimating the shape of an averaged cycle based on transition from a scalar signal to its mapping on the phase plane are shown. Original methods are proposed for estimating the dynamics of paramet...
Aim - to investigate new possibilities of application of electrocardiography (ECG) for diagnostics of myocardial ischemia with the use of modern information technologies.
Material and methods. The analyse of modern scientific sources, which were dedicated to the researches of the new uninvasion ECG-metods application was conducted. The new uninvas...
A client-server system telemedicine for remote processing of electrocardiograms was proposed. In contrast to the known, the system allows to simplify the heart activity monitoring procedure using the original ECG sensor with finger electrodes and the innovative method of ECG processing in the phase space (fasegraphy method). It is shown that the fa...
The original approach to the analysis, interpretation and economical coding of cyclic signals of a complex form is developed. The approach is based on the transition from the observed signal to the phase portrait in the multidimensional space of parameters that uniquely describe the form of the signal fragments. It is shown that the proposed approa...
The generalized generative model of artificial ECG generation with specified amplitude and time parameters of informative fragments is proposed. The model allows one to generate artificial realistic shape signals with three types of T-wave alternans with internal and external perturbations. The generated test signal can be used to assess the qualit...
We construct the generative model for creation of artificial electrocardiogram with realistic form of informative fragments. Mechanism of creation of sequence of cycles, distorted by internal and external perturbations is considered. On the basis of model experiments we reconstruct dependencies between electrocardiogram parameters in time domain an...
The formal conditions, which allow to guarantee the features utility while statistical recognition of two classes, are formulated. At the same time, only limited data on individual and general probability characteristics of the features are used. The constructability of the obtained conditions is demonstrated on the example of solving the problem o...
The problem of construction of a narrow-band rejection filter, which is based on direct and inverse discrete Fourier transform is considered. The improvement of the method is grounded theoretically due to using additional search procedure, which reduces the spectrum leakage effect. The results of testing the developed algorithm on real and model ex...
We consider an approach for estimation of utility of diagnostic tests (attributes) from the point of view of reduction of a priori risk. The statements, which provide utility of a diagnostic test and a separate attribute, are proved. Upper estimates for limit values of probabilities of errors of "target miss" and "false alarm" are obtained. It was...
The interconnection between known definitions of irrelevant and
unuseful features in statistical recognition of two classes is
considered. It is demonstrated that any relevant feature in the sense of
a posteriori probabilities change may be unuseful regarding average
error probabilities change. The necessary and sufficient conditions of
usefulness...
The concept of informative and useful features in statistical
recognition of two classes is introduced. It is demonstrated that any
feature can be informative but not useful in combination with other
features. The sufficient conditions which assume the usefulness of
feature according to error probability are consider
The paper introduces a new approach to estimat-ing a diagnostic test's utility from the point of view of risk reduction. Equations are given for the utility of a test, or separate diagnostic feature, supported by proofs. The greatest lower bound on a test's sen-sitivity and a test's specificity are obtained. Im-provement to the traditional ROC-anal...
The modified method of stationary harmonic noise filtration on the basis of discrete Fourier transformation is offered. This method is especially effective if the noise is concentrated in useful signal frequencies, when using well-known rejection filters can deform the useful signal in consequence of spectrum leakage effect. The mathematical justif...