=Paper=
{{Paper
|id=Vol-3609/paper20
|storemode=property
|title=Uncertain Model for Classification of Children's Neurological Diseases
|pdfUrl=https://ceur-ws.org/Vol-3609/paper16.pdf
|volume=Vol-3609
|authors=Olga Rybytska,Ivanna Dronyuk,Andriy Pushnyk
|dblpUrl=https://dblp.org/rec/conf/iddm/RybytskaDP23
}}
==Uncertain Model for Classification of Children's Neurological Diseases==
<pdf width="1500px">https://ceur-ws.org/Vol-3609/paper16.pdf</pdf>
<pre>
                         Uncertain Model for Classification of Children's Neurological
                         Diseases
                         Olga Rybytskaa, Ivanna Dronyuka,b and Andriy Pushnykc
                    a
                      Lviv Polytechnic National University, Bandery 12, Lviv, 79013, Ukraine
                    b
                      Jan Dlugosz University in Czestochowa, Waszyngtona 4/8, Czestochowa, 42200, Poland
                    c
                      Lviv National Medical University, Pekarska str.,69, Lviv,79010, Ukraine


                                        Abstract
                                        The dataset of thirty-three parameters about one hundred children from three to seven years
                                        old is considered. The biggest part of children (about 70%) have some neurological diseases.
                                        The dataset contains such parameters: body characteristics, physical and emotional
                                        functioning, physical health, communications, relationships in the family, and some results of
                                        medical analysis. Standard statistical analysis was realized for the preprocessing. Based on
                                        the obtained results the k-mean method was proposed for solving classification tasks. The
                                        Silhouette method and Elbow method were realized. Clusters corresponding to groups of
                                        indicators of Psychosocial Health Summary Score, Physical Health Summary Score, Total
                                        score (0-100%), and nutritional status of healthy children and children with neurological
                                        pathology were obtained. The calculation was made according to the PedsQLTM 2.0
                                        questionnaire and was calculated using a special licensed computer program Dietplan 7. An
                                        uncertain model for classification is proposed.

                                        Keywords 1
                                        Children's neurological diseases, classifications of diseases, improving diagnostics with AI,
                                        Silhouette method, Elbow method

                         1. Introduction

                            One of the most important signs of the well-being of society is the health of its members, and first
                         of all, the health of children. Modern research data indicate a tendency to decrease in children's health
                         indicators, particularly in Ukraine. Diseases showing the highest growth rates include damage to the
                         central and peripheral nervous system and primarily psychomotor disorders. The count of children’s
                         neurological diseases is increasing nowadays. The effectiveness of treatment heavily depends on the
                         timely and correct diagnosis. The development of AI technologies and their application for medicine
                         has made a real breakthrough in the diagnosis and treatment of neurological diseases in children.
                            It does not seem possible to establish the exact number (in percent of the child-age population) for
                         the purpose of comparisons by regions and the level of welfare of countries using today's existing
                         methods of diagnosis. This is due, in particular, to different possibilities in organizing samples,
                         methods, and criteria of randomization, diagnostic criteria, and instruments used. In some studies,
                         cases of pervasive developmental disorders associated with severe mental retardation, congenital
                         abnormalities of brain development, epilepsy, and prematurity are not taken into account. There is
                         often a lack of registration of children who demonstrate accelerated development, which is not
                         characteristic of their age. This explains why the prevalence of Autism Spectrum Disorders (ASD) in
                         particular in China (6.4 per 10,000) is significantly lower than in Asia, Europe, and North America

                         1                    IDDM’2023: 6th International Conference on Informatics & Data-Driven Medicine, November 17 - 19, 2023,
                         Bratislava, Slovakia
                                     EMAIL: olha.m.rybytska@lpnu.ua (A. 1); i.dronyuk@ujd.edu.pl (A. 2); apushnyk@gmail.com (A. 3)
                                     ORCID: 0000-0002-2394-355X (A. 1); 0000-0003-1667-2584 (A. 2); 0000-0001-7902-9639 (A. 3)
                                                      ©️ 2023 Copyright for this paper by its authors.
                                        Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
                                                      CEUR Workshop Proceedings (CEUR-WS.org)


CEUR
                  ceur-ws.org
Workshop      ISSN 1613-0073
Proceedings
(the average prevalence of ASD is about 1%) [1]. In South Korea, where schoolchildren were studied,
this indicator was 2.6% (3.7% among boys and 1.5% among girls) [2].
    The rate of incidence of ASD in Ukraine is increasing. Thus, from 2006 to 2017, the number of
newly diagnosed cases of the disease increased 8.5 times (by the end of 2017 it was 998) [3]. Over the
last decade, a stable trend of growth in the share of children with developmental disorders has been
recorded in Ukraine. According to information from the Center for Medical Statistics of the Ministry
of Health of Ukraine for 2017, among the children, the incidence and prevalence of central and
peripheral nervous system lesions were 40.19, respectively; 0.97 and 15.19 and 0.42 per 1000 child
population; behavioral disorders - 24.13 and 3.77 per 1000 child population.
    In particular, until 2006, ASD was almost undiagnosed in our country. The rate of incidence of
ASD in Ukraine has been increasing since 2006. The number of registered ASD patients increased
from 662 in 2005 to 7,491 in 2017. According to a large study conducted by the US Centers for
Disease Control and Prevention (CDC), the prevalence of autism spectrum disorders in the US was 1
in 88 children. The disease was five times more common in boys (1 in 54) than in girls (1 in 252).
    The prevalence of the syndrome (disorder) of attention deficit hyperactivity disorder (ADHD) is
from 4 to 12% worldwide, and of minimal brain dysfunction (MBD) - from 2 to 20%, according to
other data - from 3-8% to 30 - 50% of the child population. There is a steady trend towards an
increase in the number of children with these psychomotor disorders, as well as with mental
retardation (MR). Therefore, it can be argued that the available literature does not fully reflect the
current situation [4].
    This condition is associated with the lack of mass-planned diagnosis of psychomotor disorders at
an early age. This problem must be solved, because early diagnosis and medical, in particular, dietary
correction can significantly alleviate and sometimes eliminate symptoms. Therefore, the sooner the
disease is detected, the causes of its occurrence, as well as measures to reduce the severity of the
disease course (and in some cases complete elimination of symptoms) in children become an
extremely important task. The exact mechanisms of the effect of genetic factors on damage to the
central and peripheral nervous system are investigated with the help of genomic screening,
cytogenetic studies, and evaluation of candidate genes [5].
    Also, at the stage of primary medical care during routine preventive examinations at 9, 18, 24, and
36 months, screening of children's development is currently required, during which children with
developmental delays are identified. Additional special screening is recommended if the child has
previously described risk factors for ASD, diseases that have a high level of comorbidity with ASD.
Special screening is also recommended for older children if they have disorders of social interaction
and communication. The final diagnosis of ASD was carried out by a specialist doctor based on the
diagnostic criteria of ICD-10 (WHO, 1992), and since 2018 - ICD-11. However, the connection
between the disease and the nutritional background is mostly not considered today.
    Diseases of the central nervous system are currently considered as the result of the interaction of
genetic and environmental factors. Diseases that disrupt the work of the brain and physical influences,
particularly nutrition, are a risk of morbidity. Ensuring the quality of life is an integral part of the
treatment program for children with psychomotor disorders. In children with neurological pathology,
treatment tactics are aimed not only at alleviating the child's condition and controlling clinical
symptoms but also at improving the quality of his life [6].
    Adequate nutritional status is the key to the physiological development of the child's body, in
particular the central nervous system, in critical periods of ontogenesis (pregnancy and the first years
of life). Nutritional deficiency (excess) during these time periods leads to a violation of the formation
of cognitive functions, behavioral reactions, and productivity of the child in various spheres of his
activity, which will have a negative impact on both childhood and youth years, as well as on all
further adult life of the individual, limiting his opportunities for further development [7]. Thus, the
development and formation of the central nervous system critically depend on nutritional support that
is adequate to the child's needs in one or another period of life. Sufficient intake of proteins in the
child's body, and provision of energy substrates, including glucose, fats (long-chain polyunsaturated
fatty acids), iron, zinc, copper, iodine, and folic acid, are especially important. Their absence or
excess leads to increased metabolic disorders in the CNS, and structural and epigenetic changes in it,
which will have a long-term, often lifelong, impact on the regulation of its functions [8-12].
    Children with disorders of psychomotor development against the background of existing
neurological disorders and various somatic pathologies become especially vulnerable to deficits
(surplus) of nutritional support due to their inherent nutritional features, which significantly
complicates their rehabilitation [13]. The sensory profile of children with ASD, with its characteristic
hypo- and hyper-reactivity to tactile, olfactory, and visual stimuli, causes a strict selectivity in food
due to its appearance, consistency, taste, and temperature. [14-16]. In work [17-19] the systematic
review of the research about using modern information technologies for the diagnosis of neural
diseases. For medical diagnostics, the classification task is very important. In many modern kinds of
research, the method of linear regression [20], Nearest neighbor, Nearest Center [21], randomized
classification trees [22], and Bayesian classifier [23] are used for solving classification tasks. In work
[24,25,26] the systematic review of the research about using modern information technologies for the
diagnosis of neural diseases. For medical diagnostics, the classification task is very important. In
many modern kinds of research, the method of linear regression [27], Nearest neighbor, Nearest
Center, randomized classification trees, and Bayesian classifier [28] are used for solving classification
tasks. The recurrent approximation is used for the tasks of neural network synthesis in the medical
therapies control [29].
    Therefore, the problem of early and reliable detection of possible problems in the nutritional
supply of the child and the development of recommendations for its balancing is an important factor
in reducing the severity of the course of CNS diseases, rehabilitation, and improving the quality of life
and its expected duration.

2. Research methods

   The work uses such research methods as statistical collection of information, cluster analysis, and
construction of membership functions of a fuzzy logical model. Obtained statistical information
regarding the physical condition of children with absent central nervous system (CNS) diseases
(hereinafter "Healthy", abbreviated H) (20 children), attention deficit hyperactivity disorder (ADHD)
(13 children), mental retardation (MR) (30 children), autism spectrum disorder (ASD) (17 children),
and minimal brain dysfunction (MBD) (10 children) were systematized by age (3 years, 4-5 years, 6-7
years) and gender. The study used indicators (Group 1) of physical condition (Weight (kg), Height
(cm), Head Circumference (cm), Chest Circumference (cm), Shoulder Circumference (cm), Hip
Circumference (cm), Leg Circumference (cm), BMI (body mass index)), as well as subjective
indicators: Physical functioning, Social functioning, Emotional functioning, Functioning of
schools/kindergartens; calculated indicators: Psychosocial Health Summary Score, Physical Health
Summary Score, Total score (0-100%). The calculation was made according to the PedsQLTM 2.0
questionnaire.
   Also, in the clustering of the entire sample, the indicators were taken into account: Awareness,
Family involvement, Communication Technical skills, Emotional needs, Overall satisfaction, Total
score, Physical functioning, Emotional functioning, Social functioning, Mental functioning,
communication, Concern, Everyday activity, Relationships in the family, The Parent HRQL Summary
Score, The Family Functioning Summary.
   Average values (Group 2) of nutrients obtained from food (Water, Proteins, Fats, Carbohydrates,
Energy, Total sugar, Saturated fat, Monounsaturated fat, Polyunsaturated fat, Dietary fiber, Na, K Ca
Mg, Phosphorus (P), Fe, Cu, Zn, Cl, Mn, Se, I, Retinol, Vitamins D, E, B1, B2, PP, B6, B12, C,
Folates, Pantothenic acid, Biotin was calculated using a special of the licensed computer program
Dietplan 7, developed by the British company Forestfield Software Limited (Horsham, UK) (user
agreement dated 30.07.19).
   Clustering (k-means) was carried out in order to identify groups of similarity based on the
indicators of Group 1 and Group 2. The number of clusters [24-26] was determined according to the
elbow and silhouette methods. Initial value clusters have a large impact on the clustering model, so
algorithms are used for different given initial values. The location measure is used to find the closest
points in the clusters to the cluster center, By definition different measures of distance can be created
for different clusters. The number of clusters (k) is a defined parameter in K-Means clustering. To
determine the optimal value of various methods, in this research we use the following methods:
Elbow method; and Silhouette method.
    The Elbow method is an empirical method for determining the optimal number of clusters for a
data set. In this method, we select a subset of these k values and then apply K-Means clustering using
each k value. We find the average distance of each point in the cluster to its center. Then we calculate
the sum of squares S of the distances between the centroids. Now to determine the best number of
clusters (k), we plot k against their value S. We choose the value of k from which the plot looks like a
straight line.
    The silhouette method is unique in the way it interprets and checks for consistency in data clusters.
This method provides a concise graphical representation of how well each object is classified. The
size of the silhouette is given by a strictly mathematical formula for each cluster of points using the
minimum value of the average distance of a given point in relation to other points of the cluster.
Silhouette magnitude is a measure of how similar an object is to its own cluster (cohesion) and
different from other clusters (separation). The silhouette ranges from -1 to +1, where a high value
indicates that the object matches well with its own cluster and poorly matches neighboring clusters. If
one object has a high value, then the clustering configuration is appropriate. If many points have a low
or negative value, then the clustering configuration may have too many or too few clusters. The
silhouette can be calculated using any distance metric. In the presented research the Euclidian metric
is used.

3. Results of data clustering of patients aged 3-7 years according to all
indicators of Groups 1 and 2.

   The results of clustering the data of the entire sample into 3 clusters according to all indicators, in
particular of the 1st and 2nd groups, made it possible to confirm the fact that the composition of
nutrients is associated with diseases of the central nervous system. Thus, clusters 0 and 1 included
both healthy and sick patients. At that time, only patients with ASD, MR and ADHD were in cluster
2. The value of the centroids is shown in Fig. 1. The significant difference between the content in the
diet of such nutrients as Water, Carbohydrates, Energy, Na, K, Ca, P, Cl, Retinol, and Folates gave
the impetus for more detailed studies. Since early diagnosis and rehabilitation is the most effective, it
was decided to carry out more detailed research with data on children aged 3 years. It is obvious that
age differences played a significant role in the clustering of the sample of the 3-7 age group.


                                   0 Cluster       1 Cluster        2 Cluster

 6000
 5000
 4000
 3000
 2000
 1000
     0
                   The Family…
                    Emotional…




         Relationships in the …
                    Emotional…
           Leg circumference…


              Physical Health…
                         Head…




                       Energy




                            B2
                            B6
                   Awareness




                    Vitamin С
                             Cl
                        Water


                 Saturated fat
         Polyunsaturated fat

                            Ca

                            Cu
          Shoulder girth (cm)




                            Se

                             Е
            Emotional needs
                   Total score
                     Mass (kg)




          Mental functioning




                  Phosphorus



                       Retinol
                          Fats
                      Concern




                       Folates
         Physical functioning



             Communication




                           Na




Figure 1: The value of the centroids of clusters 0-2 of all sample participants by all indicators
4. Results of clustering of patients in the age category of 3 years.

   The results of clustering Fig. 2,3 were obtained in relation to groups of similarity according to the
indicators of Group 1 and centroid values (A group of 3-year-old patients according to the indicators
of Group 1 was clustered into 5 clusters (0-cluster - 4-cluster). Value centroid indicators for each
cluster are presented in Table 2. Both healthy and sick patients were included in clusters 0 and 3.
Patients with MR, ASD, and ADHD were included in clusters 1,2,4.




Figure 2: Results of clustering indicators of Group 1 (Emotional functioning – Social functioning)




Figure 3: Results of clustering of indicators of Group 1 (Psychosocial Health Summary Score - Physical
Health Summary Score)
   The obtained results drew attention to the non-synchronicity of the trends to increase or decrease
in the values of individual indicators of Group 1 with the presence (absence) of CNS diseases.
Moreover, when comparing the values of such an indicator as Weight in Ukrainian boys corresponds
to the norms (Tabl. 1,2), while Height and Head Circumference are deviated in the direction of
decrease from the norm. The percentage of head circumference to height corresponds to the norm. At
the same time, the breast circumference, as well as the relative index to height, exceeds the normal
values [27].
   Studies have shown that in patients with MR and ASD (to a lesser extent, ADHD), deviations in
many indicators of physical condition are extremely small and extremely large. Studies have shown
that in patients with MR and ASD (to a lesser extent, ADHD), the deviations of many indicators of
physical condition from the norm are significant, that is, their values are extremely small or extremely
large. Thus, according to the Weight indicator m, healthy children have fixed values 13.5<m<15 .5
(boys) with an average value of the norm of 14.3; and 13.5<m<17.5 (girls) with an average value of
the norm of 13.9. For patients with diseases for boys 13.5<m or 14.5<m<21.5 and for girls – m<13.5.
According to the Height indicator h, the corresponding indicators are: healthy (93.5<h<95.5 (boys),
90.5<h<94.5 (girls). For patients with diseases respectively h<93.5, or 95.5<h<105.5 (boys) and
89.5<h or 94.5<h<97.5 (girls) with an average value of the norm of 96.1 (boys) and 95.1 (girls),
respectively. A similar situation with regard to the extreme smallest and extreme largest values in
children with diagnosed psychomotor disorders is observed according to the Head Circumference,
Shoulder Circumference, Hip Circumference, and Leg Circumference indicators. Regarding the BMI
of boys, there is a deviation in critically low or critically high values, but for girls, in children with
diseases, this indicator is significantly lower than the norm.

Table 1
Values of centroids for all indicators of Group 1 (Physical and psychological health)
         Indicators of physical                                    cluster no
  development and socio-emotional             0             1           2           3                4
            functioning
               Weight (kg)                 15,66667         14,5     14,83333          16       14,71429

               Height (cm)                 94,66667        93,125    97,83333      94,83333     98,42857

         Head circumference (cm)           49,16667        49,25     49,16667          50,5     49,85714

        Chest circumference (cm)           52,83333         52,5     53,33333      54,33333     52,71429

           Shoulder girth (cm)             17,83333        17,125    17,83333      17,83333         18

         Hip circumference (cm)            27,66667        25,75         26,5          29       28,28571

         Leg circumference (cm)            22,83333         21       22,33333          23,5     22,85714

                  BMI                      17,51667     16,75125     15,49667      17,70667     15,15857

           Physical functioning            50,56667        77,425    87,51667          87       74,48571

            Social functioning                 45           80       84,16667          52,5     47,14286

          Emotional functioning                40          65,625        57,5      66,66667     64,28571

  Psychosocial Health Summary Score             9           14,2     73,56667         13,65     55,22857

Physical Health Summary Score              50,56667        77,425    87,51667          87       74,47857


   The authors believe that when building a vague model of support for the diagnosis of children with
psychomotor disorders, a significant deviation from the norm for at least two indicators of Group 1
should serve as a basis for further examination regarding the indicators of Group 2.
   In relation to Group 2 indicators of nutritional support, clustering was carried out into 3 clusters
(Fig.4-9). According to Group 2 indicators, all healthy children (except one) fell into one (2nd)
cluster.




Figure 4: Setting the number of clusters for indicators of Group 2 by the elbow method




Figure 5: Results of clustering by indicators Mn, Se. Here and in the following figures, the image is
used:
    - elements of the 0-cluster;
     - the centroid of the 0-cluster;
   - elements of the 1-cluster;
    - the centroid of the 1-cluster;
   - element and the centroid of the 2-cluster
Figure 6: Results of clustering by indicators Zn, Cl




Figure 7: Results of clustering by indicators of Fats, Carbohydrates
Figure 8: Results of clustering by indicators Vitamin C, Biotin




Figure 9: Results of clustering by indicators B12, Folate

   The obtained centroid values for all indicators of nutritional indicators are presented in Table 2
Table 2.
Values of centroids for all indicators of Group 2
                         cluster no                           cluster no                                 cluster no
Nutrient                                   Nutrient
                     0        1        2                      0        1      2    Nutrient          0        1         2
                1195,3   1107,     918,3                   783,2   3708,   519,4    Vitamin
   Water                                     Ca             6       30       2
                                                                                              1,59        0,70   1,05
                  4        70        5                                                 D
                         103,5                             233,5   573,0   155,1
  Proteins      64,42              44,36     Mg             7        0       2
                                                                                      Е       8,26        6,07   4,10
                            0
                         106,0                             1156,   3472,   786,6
    Fats        73,66              48,82   Phosphorus       28      30       6        B1      1,43        6,44   0,97
                            0
Carbohydrate    248,3    590,0     183,0
                                             Fe            9,41    70,70   6,34       B2      1,42        5,80   1,00
     s            5         0        8
                1853,    3581,     1300,                                                      13,7        46,0
   Energy                                    Cu            1,22    3,15    1,03       PP       4           5
                                                                                                                 9,47
                 61        00       24
                         151,3
 Total sugar    93,09              72,98     Zn            8,10    18,02   5,62       B6      1,19        3,35   0,89
                            0
                                                           3394,   2953,   1532,                          10,3
Saturated fat   28,13    25,63    21,07      Cl             20       0      00       B12      4,84
                                                                                                           0
                                                                                                                 3,63
Monounsatur                                                                                   171,        529,   124,
                25,72    25,80    14,61      Mn            3,12    12,15   2,34     Folates    7          00      7
   ated fat
Polyunsaturat                                                                      Pantothe               24,8
                16,43    10,00     5,73      Se            38,49   30,00   31,85              4,70
                                                                                                           0
                                                                                                                 3,49
    ed fat                                                                         nic acid
                                                           111,6                   Vitamin    67,6        342,   46,5
Dietary fiber   15,37    15,70    11,46       I                    45,30   78,71
                                                            5                         С        4          30      3
                2316,    2053,    1054,                    458,8   262,7   573,6              29,9        88,7   23,1
     Na                                    Retinol          0        0       4
                                                                                    Biotin     5           0      2
                 05       30       26
                2448,    6577,    1551,
     K
                 23       00       62
   The obtained results make it possible to obtain weighting coefficients, as well as to build
dependence functions for a fuzzy logic model [28] and AI applications [29] and support the diagnosis
and corrective diet of children with psychomotor disorders.


5. Conclusion

    The research approved that somatic health, physical development, and regulatory mechanisms
suffer in children with neurological diseases. The data obtained by testing the parents of the children
under study served as the initial data. Therefore, the indicators used for the purpose of cluster
analysis, which are calculated according to the methods specified above, are largely subjective. In
order to obtain more accurate input data, it is necessary to conduct a more thorough explanation of the
scale of each of the indicators to the persons who participate in the assessment of the value of the
studied characteristics, of all groups of indicators. Also, the authors believe that in order to reduce
subjectivity, it is necessary to separately involve objective indicators, namely the chemical
composition of blood.
    Children with disorders of psychomotor development against the background of existing
neurological disorders and various somatic pathologies become especially vulnerable to the deficiency
(surplus) of nutritional support due to their inherent nutritional characteristics. The study examined a
dataset collected by the authors. The data set was created according to the PedsQLTM 2.0
questionnaire and processed using the Dietplan 7 computer program.
    The presented study shows that the k-means clustering method can be an effective tool for better
decision-making by doctors in the diagnosis of pediatric neurological diseases. Both considered
methods can be used: silhouette and elbow. The study shows that it is advisable to divide the data set
of the first group of indicators into five clusters, and according to the group of indicators that
characterize the composition of the obtained nutrients - into three clusters according to practical
experiments. The next step of the research will be to be involved in the study of blood indicators, as
well as the history of parental diseases and the course of pregnancy with the aim of implementing
artificial intelligence methods to support the diagnosis of children's neurological diseases.
6. Acknowledgements

    The authors would like to thank the Armed Forces of Ukraine for providing security to perform
this work. This work has become possible only because of the resilience and courage of the Ukrainian
Army.

7. References

    [1] Li N., Chen G., Song X., Du W., Zheng X. Prevalence of autism-caused disability among
Chinese children: a national population-based survey, Epilepsy & Behavior. 2011. 22(4). P. 786-789.
    [2] Kim Y.S., Leventhal B.L., Koh Y.J., Fombonne E., Laska E., Lim E.C., Song D.H. Prevalence
of autism spectrum disorders in a total population sample, American Journal of Psychiatry. 2011.
168(9). P. 904-912.
    [3] Marcenkovs'kyj I.A., Marcenkovs'ka I.I. Autism spectrum disorders: risk factors, peculiarities
of diagnostics and therapy. International Neurological Journal. 2018;4(98):P. 75-83.
doi:10.22141/2224-0713.4.98.2018.139429. (In Ukrainian).
    [4] Miliutina Je.L., Kniazev V.M. Method of developing social-emotional intelligence in the
context of implementing the program of psychological support for primary school-age children with
ADHD. Psychological Research: a collection of scientific articles based on the materials of the annual
report-scientific conference of teachers, doctoral and post-graduate students of the Faculty of
Psychology "The unity of teaching and scientific research is the main principle of the university"
(May 22-26, 2023). Volnova LM, Matiash-Zaiats LP, editors. P.106-113.
    [5] Schaefer G.B., Mendelsohn N.J. Clinical genetics evaluation in identifying the etiology of
autism spectrum disorders: 2013 guideline revisions. Genetics in Medicine. 2013. 15(5). P. 399.
    [6] Niankovskyy S.L., Pushnyk А.І. Quality of life of preschool children with psychomotor
development disorders and their families. Availability of medical care and quality of its provision for
this category of patients. Child’s Health.2022; 17(2):6-18. DOI: 10.22141/2224-0551.17.2.2022.1496.
    [7] Nutrition and the developing brain, Oakville: CRC Press, 2016. 265 p.
    [8] Schipper A.L. Dietary lipid quality, environment and the developing brain. Groningen:
Rijksuniversiteit Groningen, 2018. 250 p.
    [9] Khanna D., Yalawar M., Saibaba P.V. et al. Oral nutritional supplementation improves growth
in children at malnutrition risk and with picky eating behaviors. Nutrients. 2021; 13: 3590.
doi: 10.3390/nu13103590.
    [10] Joosten K., Embleton N, Yan W, Senterre T. ESPGHAN/ESPEN/ESPR/CSPEN guidelines on
pediatric parenteral nutrition: energy. Clin Nutr. 2018; 37(6): P. 2309 - 2314.
doi:10.1016/j.clnu.2018.06.944.
    [11] Lapillonne        A., Mis Fidler          N., Goulet O., Wu J., Koletzko B.
ESPGHAN/ESPEN/ESPR/CSPEN guidelines on pediatric parenteral nutrition: Lipids.Clin Nutr.
2018; 37(6 Pt B): 2324 - 2336. doi: 10.1016/j.clnu.2018.06.946
    [12] Bronsky J, Campoy C, Braegger C. ESPGHAN/ESPEN/ESPR/CSPEN guidelines on pediatric
parenteral nutrition: Vitamins. Clin Nutr. 2018; 37: P. 2366-2378. doi: 10.1016/j.clnu.2018.06.951.
    [13] Pushnyk A.I., Nyankovskyy S.L. Nutrient support for preschool children with psychomotor
disorders. Child’s health. 2023;18(1):35-43, doi: 10.22141/2224-0551.18.1.2023.1555.
    [14] Bjørklund G., Waly M. I., Al-Farsi Y. et al. The role of vitamins in autism spectrum disorder:
What do we know?. J. Mol. Neurosci. 2019. Vol. 67(3). Р. 373–387. doi:10.1007/s12031-018-1237-5.
    [15] Esteban-Figuerola P., Canals J., Fernández-Cao J.C., Arija V. (2019). Differences in food
consumption and nutritional intake between children with autism spectrum disorders and typically
developing       children:     A     meta-analysis.     Autism.2019.,      Vol.23(5).     Р.1079–1095.
doi:10.1177/1362361318794179.
    [16] Abbasi K., Beigrezai S., Ghiasvand R. et al. Dietary patterns and Attention Hyperactivity
disorder among Iranian children: a case-control study. J American Сoll Nutr. 2019. №38(1). P. 76-83
doi:10.1080/07315724.2018.1473819.
    [17] Previously Marzena Szkodo MOR, Micai M, Caruso A, Fulceri F, Fazio M, Scattoni M.L.
Technologies to support the diagnosis and/or treatment of neurodevelopmental disorders: A
systematic       review.     Neurosci      Biobehav     Rev.       2023      Feb;145:105021.     doi:
10.1016/j.neubiorev.2022.105021.
    [18] Bondioli M., Chessa S., Kocian A., Pelagatti S., (2023) A survey on technological tools and
systems for diagnosis and therapy of autism spectrum disorder, Human-Computer Interaction,
DOI: 10.1080/07370024.2023.2242345
    [19] S. Pandey and S. Sharma, ML Algorithms that have been Utilised to Classify Neuro-
Developmental Disorders: A Review, 2023 6th International Conference on Information Systems and
Computer        Networks       (ISCON),      Mathura,      India,      2023,     pp.    1-5,     doi:
10.1109/ISCON57294.2023.10112157
    [20] Stefano Benati, Sergio García, A mixed integer linear model for clustering with variable
selection,     Computers      &     Operations     Research,      Vol.    43,     2014,   P.280-285,
https://doi.org/10.1016/j.cor.2013.10.005.
    [21] Yifeng Zhao, Liming Yang, Distance metric learning based on the class center and nearest
neighbor       relationship,     Neural     Networks,     Volume        164,     2023,    P.631-644,
https://doi.org/10.1016/j.neunet.2023.05.004.
    [22] Rafael Blanquero, Emilio Carrizosa, Cristina Molero-Río, Dolores Romero Morales, Optimal
randomized classification trees, Computers & Operations Research, Vol. 132, 2021, 105281,
https://doi.org/10.1016/j.cor.2021.105281.
    [23] Rafael Blanquero, Emilio Carrizosa, Pepa Ramírez-Cobo, M. Remedios Sillero-Denamiel,
Variable selection for Naïve Bayes classification, Computers & Operations Research, Volume 135,
2021, 105456, https://doi.org/10.1016/j.cor.2021.105456.
    [24] Rybytska O. M., Levchenko O., Dilai M. A. value approach to forming a fuzzy model for
evaluating business models of IT enterprises in Ukraine. CEUR Workshop Proc. – 2023. – Vol. 3403.
Vol. III: P. 168–178.
    [25] Shpak N., Karpyak A., Rybytska O., Gvozd M., Sroka W. Assessing the business models of
Ukrainian IT companies // Forum Scientiae Oeconomia. – 2023. – Vol. 11, No. 1. – P. 13–48.
    [26] Karpyak A., Rybytska O. Cluster analysis of motivational management of personnel support
of IT companies. CEUR Workshop Proceedings. 2022. Vol. 3171(1), P. 1684–1693.
    [27] World Health Organization. Child growth standards, https://www.who.int/tools/child-growth-
standards
    [28] Zaiats V.M., Rybytska O.M., Zaiats M.M., An Approach to Assessment of the Value and
Quantity of Information in Queueing Systems Based on Pattern Recognition and Fuzzy Sets Theories.
Cybernetics and Systems Analysis. 2019. V. 55/4. P.638-648. https://doi.org/10.1007/s10559-019-
00172-1
    [29] Trunov A. Recurrent approximation in the tasks of the neural network synthesis for the
control of the process of phototherapy. Computer Systems for Healthcare and Medicine (2017), P. 213
– 248

</pre>