<!DOCTYPE article PUBLIC "-//NLM//DTD JATS (Z39.96) Journal Archiving and Interchange DTD v1.0 20120330//EN" "JATS-archivearticle1.dtd">
<article xmlns:xlink="http://www.w3.org/1999/xlink">
  <front>
    <journal-meta />
    <article-meta>
      <title-group>
        <article-title>Information-Analytical Support to Medical Industry</article-title>
      </title-group>
      <contrib-group>
        <contrib contrib-type="author">
          <string-name>n Tsmots</string-name>
          <email>ivan.tsmots@gmail.com</email>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <contrib contrib-type="author">
          <string-name>syl T</string-name>
        </contrib>
        <contrib contrib-type="author">
          <string-name>slyuk</string-name>
          <xref ref-type="aff" rid="aff0">0</xref>
        </contrib>
        <aff id="aff0">
          <label>0</label>
          <institution>Department of Electronic Business, University of Vienna</institution>
          ,
          <addr-line>Vienna</addr-line>
          ,
          <country country="AT">AUSTRIA</country>
        </aff>
      </contrib-group>
      <fpage>0000</fpage>
      <lpage>0002</lpage>
      <abstract>
        <p>The list of tasks that are solved by the medical information-analytical system is determined and it is shown that it is expedient to develop such a system based on an integrated approach, covering information technologies, methods, and tools for data analytics, modeling, forecasting and decision-making. It is suggested to build a medical information-analytical system on the principles of systematicity, variable equipment composition, modularity, openness, compatibility and use of a set of basic design solutions. The component-hierarchical design method has been improved, the architecture of the medical informationanalytical system has been developed, and the list of components and tasks to be solved has been defined.</p>
      </abstract>
      <kwd-group>
        <kwd>Medical Information Technology</kwd>
        <kwd>Information-Analytical Systems</kwd>
        <kwd>Component-Hierarchical Method</kwd>
        <kwd>Neural Network</kwd>
        <kwd>Data Mining</kwd>
      </kwd-group>
    </article-meta>
  </front>
  <body>
    <sec id="sec-1">
      <title>-</title>
      <p>MIT depends to a large extent on the formalization of medical (diagnostic)
information achieved by the collaboration of developers and physicians.</p>
      <p>The widespread adoption of information technology in the healthcare industry
requires the development of MIT components that simply adapt to the requirements of
specific applications, focused on the synthesis of a wide range of medical information
and information-analytical systems (IAS). The usage of pre-made components for the
synthesis of IAS for specific applications will reduce the time and cost of their creation.
The main task of implementing information technology in medicine is to improve the
level and effectiveness of medical care. Such implementation is linked to the creation of
hierarchical multilevel information systems (regional, territorial, local and individual),
which should ensure effective interaction between system levels through the exchange
of information in the form of information flows. Streamlining information flows is
needed to improve the organizational structure of health care management, optimize
health control processes, improve the workflow, and automate the process of obtaining,
collecting, storing, searching, and processing data. Streamlining information flows at all
hierarchical levels will increase the efficiency of the health care system and ensure the
economical usage of human, financial and material resources.</p>
      <p>With this in mind, the urgent task is to develop and use information-analytical tools
to improve the effectiveness of treatment processes.
2</p>
    </sec>
    <sec id="sec-2">
      <title>Analysis of recent research and publications</title>
      <p>In recent years, there have been many studies and publications devoted to the issues
of introduction the information technologies in the medical field, development of
hardware and software components and synthesis on their basis of medical IAS
[111]. Have a look at the most important of these publications. One of the main areas of
development the medical information technologies is computerization of medical
equipment, automation of collecting information process on the patient's condition,
real-time processing of data and usage of the results of such processing to support
decision-making. The disadvantage of existing medical information systems is the
lack of intelligent means of processing and analysis of tomography data, ultrasound
diagnostics, electroencephalography, electrocardiograms and radiography.</p>
      <p>
        In works [
        <xref ref-type="bibr" rid="ref1 ref2 ref3 ref4 ref5">1-5</xref>
        ] the principles of construction and architecture are analyzed, the
main of which is modularity of construction, openness and compatibility. Existing
medical information systems do not allow the prompt storage and editing of large
amounts of data and knowledge. In most cases, storing and preprocessing data comes
down to the three main steps of placing, sorting, and finding need data. The
development of specialized medical components of data collection and processing that
provide a new level of instrumental research methods is analyzed.
      </p>
      <p>
        The details of medical information system software development have been
analyzed in publications [
        <xref ref-type="bibr" rid="ref3 ref4 ref5 ref6">3-6</xref>
        ]. It is shown that the software application is based on
mathematical methods of biomedical data processing, and the development of algorithms
is performed jointly with the doctors of the relevant specialties, who are task-makers.
The disadvantage of the software components of existing medical information
systems is that they do not fully cover the entire research process, which includes the
steps of preparation, research and data processing.
      </p>
      <p>From the analysis follows, that when designing modern medical IAS, it is
necessary to widely use ready-made components, to create a single information space with
reliable and complete information. It is also advisable to use integration as a way of
organizing individual components into one system, which ensures a coherent and
purposeful interaction between them, which results in high efficiency of the whole
system. Integration in medical IAS is carried out in functional, informational,
software-algorithmic and technical directions.</p>
      <p>The purpose of work – choice of development principles, improvement of
component-hierarchical method of designing and development of medical IAS. To achieve
this goal must be performed the following tasks:
 identification of the tasks to be solved and choice of development principles of
medical IAS;
 improvement of component-hierarchical design method of medical IAS;
 development of architecture of medical information-analytical system;
 determining the list of components and tasks that they solve.
3</p>
    </sec>
    <sec id="sec-3">
      <title>Identification of the tasks to be solved and choice of development principles of medical IAS</title>
      <p>It is proposed the development of medical IAS with an integrated approach based on
new computer information technologies and covers information, organizational,
technical, software and mathematical software. Information technology for the medical
industry is closely linked to the accumulation and processing of information. The
accuracy and reliability of diagnosis, as well as the effectiveness of treatment, depends
largely on the methods of working with medical information. When working with
information, the diagnosis is based on the information about the patient's running state and
their interpretation. In order to accomplish such tasks, IAS components should:
 data collection, archiving and preliminary evaluation;
 generating reports on all necessary parameters of a medical card, as well as having
the means to create report templates;
 automated data entry, using predictive text input and auto-fill fields;
 error detection, correctness control of entered data, monitoring of data’s entry
timeline, detection of unexpected or dangerous assignments, etc.
 protection of medical data from unauthorized access;
 prompt and intelligent processing of medical data, modeling;
 forecasting and planning of events of different medical nature;
 multidimensional data visualization, presentation of medical data.</p>
      <p>In order to reduce the cost, timing and expansion of application areas, MIT is
proposed to be implemented using the following principles:
 the systematic way in which links are formed between the IT components that
provide integrity and interoperability with other systems;
 variable composition of equipment, which implies the presence of a kernel of
information technology and variable software and hardware modules, by which the
kernel adapts to the requirements of a specific application;
 modularity, which involves the development of components of medical
information technology in the form of functionally completed modules that have access
to a standard interface;
 openness in which medical information technologies are created taking into
account the possibility of replenishing and updating functions without disturbing
their functioning;
 compatibility that involves the usage of IT interfaces through which medical
information technologies can interact with other systems;
 consistency of data intensities with the computing power of hardware;
 usage of a set of basic design solutions.</p>
    </sec>
    <sec id="sec-4">
      <title>Component-hierarchical method of designing medical information-analytical system</title>
      <p>The development of medical IAS is advisable to perform on the basis of the
componenthierarchical method, which involves the division of the development process into
hierarchical levels and types of support (algorithmic, hardware and software). To implement
this approach, the decomposition method is used, which involves splitting MIT into
individual components. At each level of the hierarchy, tasks of corresponding
complexity are solved, characterized by both units of information and processing algorithms. The
complexity of the tasks is divided into four hierarchical levels. Increasing the number of
hierarchy levels corresponds to the increasing detail of algorithmic, hardware and
software tools. At the higher levels of the hierarchy, information units, algorithms, software,
and hardware are ordered sets of information units and compositions of algorithms,
software, and hardware of the lower levels of the hierarchy (Table 1). The sequential
decomposition methodology used in the development of medical information
technology media reflects the top-down development process.
At the first hierarchical level of IAS development, system tasks are solved. We denote
1
this level as CМІТ , where the unit means the first level of the hierarchy. The second
level of the hierarchy consists of subsystems that provide: registration and collection
of medical data; processing of medical data; data retention; analytical processing of
medical data; decision-making. The third hierarchical level consists of hardware
blocks and libraries, which are divided by functional features. The fourth level
includes hardware and software modules that implement the basic algorithms of blocks.</p>
      <p>The component-hierarchical structure of MIT can be described by equation:
CМ1ІТ  nC М2iІТ m C М3ІjТ hC М4ІpТ ,</p>
      <p>i1 j1 p1
where C М2iІТ , C М3jІТ , C М4pІТ - means of the second, third and fourth hierarchical levels
respectively; n – number of subsystem types; m - number of block types; h - number
of module types.</p>
      <p>The algorithms of the hardware at each level of the hierarchy are represented as
functional graphs. F=(Ф,Г), where Ф={Ф1,Ф2,…,Фn} – set of functional operators,
Г- the law of displaying links between operators. Such representation allows us to
create the list of algorithms and to determine expediency of their implementation by
software or hardware.</p>
      <p>In the treatment course of a patient by medical professionals a large amount of
various information is created (text records, spreadsheets, graphics), which is processed
by information technologies that combine into one integrated medical technology.
Therefore, the MIT architecture under development can be regarded as an integrated
technology for processing information and the means used to implement it:
ІТ МІТ  ІТ DW ,ІТWEB , ІТ ЦОС , ІТ OLAP , ІТ EDMS , ІТ DM , ІТ KDD ,
where ІТDW – technology of information repositories (Data Warehouse); ІТWEB –
WEB-technology; ІТЦОС – digital signal processing (DSP) technologies; ІТOLAP –
technology of prompt analytical processing (OLAP - On-Line Analytical Processing);
ІТEDMS – technology of business process automation (EDMS - Enterprise Document
Management System); ІТDM – data mining technology (DM - Data Mining); ІТKDD –
technology that extracts new non-trivial knowledge from data in the form of models,
dependencies and laws (KDD - knowledge discovery in databases).</p>
      <p>The main role of integrated medical technology is to scientifically substantiate and
find new solutions at the intersection of formal and logical approaches, taking into
account the empirically descriptive nature of medicine.
5</p>
    </sec>
    <sec id="sec-5">
      <title>Architecture and components of the medical IAS</title>
      <p>The development of medical IAS is proposed to be carried out using a systematic
approach that covers all levels of integration of processes and hardware. Synthesis of
medical IAS is advisable to carry out on the basis of ready hardware and software
since the development and production of new ones require considerable money and
time. When choosing components, there are many factors to be considered, namely:
information about the finished hardware and software components, their technical
characteristics, compliance of the interfaces with standards, opportunities for their
purchase, and more. The trend of medical IAS development is focused on the
transformation from a convenient tool for collecting, storing and processing data to a tool
for supporting effective management decisions. For information-analytical support of
treatment processes, the architecture of medical IAS was developed (Fig. 1), where
DBMS - database management systems.
The main components of the developed medical IAS are: subsystems of registration and
medical data collection; medical data processing subsystem; data storage subsystem;
subsystem of analytical processing of medical data; the decision support subsystem.</p>
      <p>Subsystem for registration, collection and preprocessing of medical data. The
main component of medical information systems at all levels is the subsystem of
collection, pre-processing and storage of medical data. Such subsystem shall provide:
formalization, filtering and sorting of data; automated entry of paper documents into
electronic form; registration, accounting of all volume of incoming, outgoing and
internal documents; primary processing and registration of documents, data entry in
databases; prompt search of documents and search of documents according to the
request by attributes of the document (registration number, date, authors, performers,
etc.), keywords and descriptions of document fragments; optimal usage and
systematization of data warehouses according to information technology needs; integration
and interaction with Web-technologies, e-mail and file systems; support of various
sources of information receipt; possibility to work with modern DBMS.</p>
      <p>The structure of the subsystem for collection, pre-processing and storage of
medical data is shown in Fig. 2, where ODS is an operational data source.
The main components of the medical data collection and storage subsystem are data
classification tools, data structure enhancement, text parsing, data preparation, and
operational data sources. It worse having a closer look at the processes that precede
the recording of information in data warehouses.</p>
      <p>Means of medical data registration. Much of the medical data is recorded in a
variety of text documents (referrals for research, analysis, medical history, prescription,
activity report), images (CT, X-rays, ultrasound) and digital signals (cardiographs).</p>
      <p>Throughout his work, a medical professional fills in the appropriate standard forms
of medical records in which. the following data are recorded:
 patient information - surname, first name, patronymic, year and place of birth,
nature of work, relatives;
 data on the course of treatment - the results of laboratory, instrumental studies,
prescribed treatment, the impact of medications, medical findings;
 data on the structure, function of medical institutions, laboratory and instrumental
methods of research;
 statistical-management data (indicators of diagnosis accuracy (according to WHO
classification), length of staying in hospital, degree of recovery, disagreement in
diagnoses), used to calculate state medical statistics for institutions and indicators
that characterize the work of a physician, department and institution as a whole;
 economic-planned data on practical and accounting activity of medical institutions.
Medical data can be obtained in the following ways:
 from analog media by recognizing the text of medical documents, digitizing
images and audio data;
 directly entering data into electronic forms or from the outputs of computer
tomographs, x-rays, ultrasound scanners, and cardiographs.</p>
      <p>Obtained digital medical data is being formalized and filtered before storage. Medical
data can be divided into three classes:
 highly structured (directions, prescriptions, laboratory analysis results) - have a
given document structure, clear formats and filling rules;
 partially structured data (some rules and formats defined but in very generalized
form) - results of instrumental studies and medical prescriptions;
 unstructured data (information written in arbitrary text) - physicians' findings,
descriptions of procedures and other information.</p>
      <p>Means of data structure improving. One of the most important conditions that ensures
the efficiency of medical information processing is its formalization, in which data
obtained from different sources is reduced to one form. In order to identify certain
trends and patterns, it is advisable to group and submit the medical data in tabular
form. Obviously, the mechanisms for preparing, filtering, analyzing, and processing
highly structured data are far more productive than arbitrary text analysis methods.
Therefore, to effectively use medical information technology, it is necessary to
increase the level of data structure.</p>
      <p>In MIT entering of medical information is done using medical cards as well as
modern medical devices (computer tomographs, fluorographs, ultrasound scanners,
cardiographs, etc.), that shape the results of work electronically. The registration and
collection subsystem of medical data must ensure:
 automated entring of paper documents into electronic form;
 registration, accounting of all volume of incoming, outgoing and internal documents;
 primary processing and registration of documents, data entry into a database;
 prompt document search according to the document attribute request (registration
number, date, authors, performers, etc.), keywords and document snippets;
 optimal usage and systematization of data warehouses according to MIT needs;
 integration and interaction with Web-technologies, e-mail and file systems;
 support for various sources of information;
 possibility to work with modern DBMS.</p>
      <p>One of the goals of MIT creating is to improve care delivery, no matter where a
patient is. This is accomplished through the provision of timely and targeted
highquality healthcare through telecommunications and computer technology combined
with the experience of medical specialists (telemedicine).</p>
      <p>The medical data storage subsystem is focused on the secure storage of large
amounts of data, which uses data storage, that consists of two main parts: database
and DBMS. Data in the subsystem can be stored in different databases and when
analyzing them, there may be problems with supporting of different data formats, as well
as with their encoding. This problem is solved by creating data warehouses that are
domain-oriented, integrated, and immutable that support the history of the dataset.</p>
      <p>The MIT databases contain the following: patient history; relocation of patients to
a specialized unit; appointment of additional methods of medical examination,
laboratory analyzes and referrals, as well as surgical intervention.</p>
      <p>Knowledge bases used in MIT are filled with data on the experience of
problemsolving by doctors in hospitals, specialized clinics, and rehabilitation centers. The
experience of doctors, which is introduced into the knowledge base, is a verbal
algorithm of diagnostic, therapeutic, prophylactic and rehabilitation measures with an
assessment of their effectiveness.</p>
      <p>Analytical processing of the medical data subsystem uses the following tools:
OnLine Analytical Processing (OLAP), Data Mining (DM) and modeling.</p>
      <p>
        Methods and tools for analytical data processing are based on the usage of classical
statistical approaches, averaged indicators, on the basis of which to test the
preformulated hypotheses and "rough" intelligence analysis. Standard static methods
reject atypical observations - so-called peaks and bursts. Although such information
may be interest in studying, characterizing some important phenomena. The analysis
and detailed examination of such observations is useful for understanding the nature
of the objects or investigated phenomena [
        <xref ref-type="bibr" rid="ref10 ref11 ref7 ref8 ref9">7-11</xref>
        ].
      </p>
      <p>
        The main tasks of the IAD are: searching for functional and logical patterns in the
accumulated data; finding hidden rules and patterns; building models and rules that
characterize a state or predict development of certain processes. In general, the IAD
process is divided into such stages [
        <xref ref-type="bibr" rid="ref10">10</xref>
        ]:
 pattern detection (free search);
 usage of identified patterns to predict unknown values (predictive modeling);
 exception analysis designed to detect and interpret anomalies in the patterns found.
Depending on the principles of working with the initial training data, all IAD methods
can be divided into two large groups:
 based on the direct usage of training data;
 based on the usage of data extracted from primary data and transformed into some
formal constructs.
      </p>
      <p>The first group methods are used in the stages of predictive modeling and exception
analysis. This group of methods includes: cluster analysis, nearest neighbor method,
k-nearest neighbors method, judgment by analogy. The second group of methods
includes: logical methods (genetic algorithms, decision trees, fuzzy queries and
analyzes, character rules); methods of visualization; cross-tabulation methods (agents,
Bayesian networks, cross-tabular visualization); equation-based methods (statistical
methods and neural networks).</p>
      <p>Modeling tools are an important means of solving many problems, in particular,
modeling a functional state of a patient, his organs and systems. Using models at MIT
reveals useful information that is used to make decisions. Models can be written in the
form of diagrams, mathematical formulas, etc. The simplest formal description of a
model is made with the following functional dependence:</p>
      <p>Y  f (x1,..., xn , z1,...,z m ,w1,...,wk ) ,
where Y – dependent or target variable x1,…,xn – independent variables that are
internal characteristics of an object of study; z1,…,zm - independent variables, which are
external factors that affect the object of study; w1,…,wk – not taken into account
characteristics or factors.</p>
      <p>The advantage of using models to diagnose and investigate the treatment process is
the simplicity of a model compared to a real object. These models allow to identify
the most important factors in the object in terms of study purpose. Developed models
can have different complexity, which depends on both - the complexity of a diagnosis
and used treatments.</p>
      <p>The models used in MIT can be classified according to the characteristics of the
object being investigated as follows:
 dynamic and static;
 stochastic and deterministic;
 continuous and discrete;
 linear and nonlinear;
 statistical, expert, based on Data Mining methods;
 predictive, classification and descriptive.</p>
      <p>
        When several different models have been developed, their selection is based on
characteristics and estimates, as well as expert opinion. The main characteristics of a model
that are taken into account when choosing the model are the accuracy and efficiency of
the algorithm. After testing, evaluating and selecting the model, the application stage
comes. At this point, the selected model is used for new data to solve tasks. The
decision-making problem in information MIT has the following common features [
        <xref ref-type="bibr" rid="ref11">11</xref>
        ]:
 uniqueness of choice situation;
 сdifficult to evaluate character alternatives;
 uncertainty of the aftereffects;
 many different factors to consider when making decisions;
 the presence of a person or group of decision-makers.
      </p>
      <p>In MIT, the decision-making subsystem performs the following functions:
 helps the doctor to evaluate the situation, select criteria and evaluate their relative
importance;
 generates possible solutions and scenarios;
 assesses and selects solutions and scenarios;
 ensures constant exchange and coordination of information on the progress of the
decision-making process;
 modeling and analysis of the possible consequences of the decisions made;
 evaluates the implementation of made decisions and conducts further training on
their results.</p>
      <p>The decisions that doctors need to make are very diverse. At one level, decisions
change from “structured” to “unstructured”. Structured decisions are repetitive, in
accordance with the established order, decisions for which some procedure has been
developed. Unstructured solutions, on the contrary, are solutions that are not in order
and even unique. There are no pre-agreed procedures for making such decisions and
the decision-maker must be guided by own judgments, estimations and situation.
Many of the solutions are between these two groups and are called “semi-structured
solutions”, which means that only part of the problem can be answered clearly
through a pre-agreed procedure.</p>
      <p>At this level, when making sufficiently structured and predictable decisions, some
technologies that collect, store, and facilitate access to operational-level information
may be particularly useful. Operational Processing Systems (OPS) are the basic
systems for operating at the operational level. They process thousands of common
operations that occur daily in most organizations. In addition to the fact that these systems
update records and documents in accordance with new data, they often produce
convenient short reports. The information contained in such systems at the operational
level is an important basis for higher-level systems.
6</p>
      <p>Conclusion
1. It is advisable to carry out component development and synthesis of the IAS
architecture on the basis of an integrated approach that covers information technologies,
methods, and tools of analytical data processing, modeling, forecasting and
decision making, and is based on the following principles of construction: systematic,
variable equipment, modularity, openness, compatibility and usage of basic design
solutions set.</p>
    </sec>
  </body>
  <back>
    <ref-list>
      <ref id="ref1">
        <mixed-citation>
          1.
          <string-name>
            <surname>Hrytsenko</surname>
            ,
            <given-names>V.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Kotova</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Vovk</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Kiforenko</surname>
            ,
            <given-names>S.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Belov</surname>
            ,
            <given-names>V.</given-names>
          </string-name>
          :
          <source>Information Technology in Biology and Medicine: Lecture Course: Tutorial</source>
          , Naukova Dumka,
          <string-name>
            <surname>Kyiv</surname>
          </string-name>
          (
          <year>2007</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref2">
        <mixed-citation>
          2.
          <string-name>
            <surname>Korolyuk</surname>
            ,
            <given-names>I.</given-names>
          </string-name>
          : Medical informatics: Textbook, Ofort, Samara (
          <year>2012</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref3">
        <mixed-citation>
          3.
          <string-name>
            <surname>Medykovskiy</surname>
            ,
            <given-names>M.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Tkachenko</surname>
            ,
            <given-names>R.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Tsmots</surname>
            ,
            <given-names>I.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Tsymbal</surname>
            ,
            <given-names>Yu.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Doroshenko</surname>
            ,
            <given-names>A.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Skorohoda</surname>
            ,
            <given-names>O.</given-names>
          </string-name>
          :
          <article-title>Intelligent components of integrated automated control systems: monograph, Publisher of Lviv Polytechnic</article-title>
          ,
          <string-name>
            <surname>Lviv</surname>
          </string-name>
          (
          <year>2015</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref4">
        <mixed-citation>
          4.
          <string-name>
            <surname>Berezsky</surname>
            ,
            <given-names>O.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Batko</surname>
            ,
            <given-names>Yu.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Berezska</surname>
            ,
            <given-names>K.</given-names>
          </string-name>
          :
          <article-title>Methods, algorithms and software for biomedical image processing: monograph, Ekonomichna dumka TNEU</article-title>
          , Ternopil, (
          <year>2017</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref5">
        <mixed-citation>
          5.
          <string-name>
            <surname>Dubchak</surname>
            ,
            <given-names>L.</given-names>
          </string-name>
          :
          <article-title>Telemedicine: current state and prospects of development</article-title>
          .
          <source>In Systems of information processing</source>
          , vol.
          <volume>1</volume>
          , pp.
          <fpage>144</fpage>
          -
          <lpage>146</lpage>
          (
          <year>2017</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref6">
        <mixed-citation>
          6.
          <string-name>
            <surname>Rashkevych</surname>
          </string-name>
          , Yu.,
          <string-name>
            <surname>Tkachenko</surname>
            ,
            <given-names>R.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Tsmots</surname>
            ,
            <given-names>I.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Peleshko</surname>
            ,
            <given-names>D.</given-names>
          </string-name>
          :
          <article-title>Neural-Like methods, algorithms and structures for real-time signal and image processing: monograph, Publisher of Lviv Polytechnic</article-title>
          ,
          <string-name>
            <surname>Lviv</surname>
          </string-name>
          (
          <year>2014</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref7">
        <mixed-citation>
          7.
          <string-name>
            <surname>Berezsky</surname>
            ,
            <given-names>O.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Pitsun</surname>
            ,
            <given-names>O.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Batryn</surname>
            ,
            <given-names>N.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Datsko</surname>
            ,
            <given-names>T.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Berezska</surname>
            ,
            <given-names>K.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Dubchak</surname>
            ,
            <given-names>L.</given-names>
          </string-name>
          :
          <article-title>Modern automated microscopy systems in oncology</article-title>
          .
          <source>In: Proceedings of the 1st International Workshop on Informatics &amp; Data-Driven Medicine</source>
          , pp.
          <fpage>311</fpage>
          -
          <lpage>325</lpage>
          (
          <year>2018</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref8">
        <mixed-citation>
          8.
          <string-name>
            <surname>Berezsky</surname>
            ,
            <given-names>O.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Verbovyy</surname>
            ,
            <given-names>S.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Pitsun</surname>
            ,
            <given-names>O.</given-names>
          </string-name>
          :
          <article-title>Hybrid Intelligent information technology for biomedical image processing</article-title>
          .
          <source>In: Proceedings of the IEEE International Conference Computer Science and Information Technologies (CSIT'2018)</source>
          , pp.
          <fpage>420</fpage>
          -
          <lpage>423</lpage>
          (
          <year>2018</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref9">
        <mixed-citation>
          9.
          <string-name>
            <surname>Tsmots</surname>
            ,
            <given-names>I.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Teslyuk</surname>
            ,
            <given-names>V.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Teslyuk</surname>
            ,
            <given-names>T.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Ihnatyev</surname>
            ,
            <given-names>I.</given-names>
          </string-name>
          <article-title>Basic components of neuronetworks with parallel vertical group data real-time processing</article-title>
          .
          <source>Advances in Intelligent Systems and Computing II, Advances in Intelligent Systems and Computing 689</source>
          , pp.
          <fpage>558</fpage>
          -
          <lpage>576</lpage>
          , Springer (
          <year>2018</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref10">
        <mixed-citation>
          10.
          <string-name>
            <surname>Khavalko</surname>
            ,
            <given-names>V.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Tsmots</surname>
            ,
            <given-names>I.</given-names>
          </string-name>
          :
          <article-title>Image classification and recognition on the base of autoassociative neural network usage</article-title>
          .
          <source>In: Proceedings of the IEEE 2nd Ukraine Conference on Electrical and Computer</source>
          Engineering (UKRCON-
          <year>2019</year>
          ), pp.
          <fpage>1118</fpage>
          -
          <lpage>1121</lpage>
          (
          <year>2019</year>
          ).
        </mixed-citation>
      </ref>
      <ref id="ref11">
        <mixed-citation>
          11.
          <string-name>
            <surname>Tkachenko</surname>
            ,
            <given-names>R.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Izonin</surname>
            ,
            <given-names>I.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Vitynskyi</surname>
            ,
            <given-names>P.</given-names>
          </string-name>
          ,
          <string-name>
            <surname>Lotoshynska</surname>
            ,
            <given-names>N.</given-names>
          </string-name>
          , and
          <string-name>
            <surname>Pavlyuk</surname>
            ,
            <given-names>O.</given-names>
          </string-name>
          :
          <article-title>Development of the Non-Iterative Supervised Learning Predictor Based on the Ito Decomposition and SGTM Neural-Like Structure for Managing Medical Insurance Costs</article-title>
          .
          <source>In Data</source>
          , vol.
          <volume>3</volume>
          , no.
          <issue>4</issue>
          , pp.
          <fpage>1</fpage>
          -
          <lpage>14</lpage>
          (
          <year>2018</year>
          ).
        </mixed-citation>
      </ref>
    </ref-list>
  </back>
</article>