Machine Learning Possibilities for Evaluation of Arterial

Hypertension Treatment Efficiency in Case Study

Vladimir S. Kublanov

, Yan E. Kazakov

and Anton Yu. Dolganov

Ural Federal University, Yekaterinburg, Russian Federation

“Medical Technologies” JSC, Yekaterinburg, Russian Federation

Keywords: Arterial Hypertension, Heart Rate Variability, Machine Learning, Quadratic Discriminant Analysis.

Abstract: The paper aims to discuss questions concerning application of the machine learning based decisions in the

area of the clinical diagnostics. In previous works it was shown that it is possible to develop a decision support

system based on the most indicative parameters of the short-term heart rate variability signals for the express

diagnosing of the arterial hypertension using methods of machine learning. This paper show results of the

case-study for analysis of the machine learning based results for evaluating duration of the treatment using

the device for the neuro-electrostimulation. Comparative analysis of the results of the quadratic discriminant

analysis application and instrumental measurements highlights concern regarding using of a single method in

such complex task as a clinical process. Possible limitations and advantages of each method were discussed.

1 INTRODUCTION

It is known that the functional purpose of the

autonomic nervous system (ANS) is to maintain a

constant internal environment (homeostasis) and

provide various forms of mental and physical activity.

Autonomic disorders are one of the urgent problems

of modern medicine. This is due to several factors,

and above all, the enormous prevalence of autonomic

disturbances (up to 80% of observations occur). With

violations of the ANS, fluctuations in blood pressure

are observed in 36% of patients. Arterial hypertension

was chosen as a clinical model accompanied by

autonomic disorders (da Silva et al., 2014).

An analysis of the pathophysiological factors of

arterial hypertension (AH) indicates the exceptional

role of the ANS in the formation of AH (Kseneva,

Borodulina, Trifonova, & Udut, 2016). One indirect

way to assess functioning of the ANS is the

evaluation of heart rate variability (HRV) (Baevskiy,

2001). A number of studies claim that the prognostic

significance of HRV itself is very moderate.

However, in combination with other methods, it

becomes even more significant in assessing cardiac

mortality and rhythm disturbances. Evaluation of the

https://orcid.org/0000-0001-6584-4544

https://orcid.org/0000-0002-2332-6826

https://orcid.org/0000-0003-2318-9144

results of functional tests requires special attention, as

their use has serious advantages, since it allows one

to minimize individual differences and evaluate the

direction of changes (Ushakov, Orlov, Baevskii,

Bersenev, & Chernikova, 2013).

In the treatment of hypertension can be used as

pharmacological agents and physiotherapy

techniques. According to guidelines for hypertension

Russian Medical Society of arterial hypertension and

Russian Scientific Society of Cardiology, for the

treatment of hypertension used a combination of

several drugs.

Adverse events in this case may be one of the

possible side effects is polypharmacy. Methods of

physiotherapy effects can be used both independently

and accompanying with a pharmacological drug.

Thus during the treatment process usually provided

correction of autonomic disturbances aimed to

improving the blood circulation system (Mancia et

al., 2013). However, far from always in this case the

required clinical effect required clinical effect of

optimizing the state of the ANS is provided.

In the present work, the results of clinical trials in

the treatment of patients with AH are presented. In

these studies, used device allowed for application in

Kublanov, V., Kazakov, Y. and Dolganov, A.

Machine Learning Possibilities for Evaluation of Arterial Hypertension Treatment Efﬁciency in Case Study.

DOI: 10.5220/0009372004110416

In Proceedings of the 13th International Joint Conference on Biomedical Engineering Systems and Technologies (BIOSTEC 2020) - Volume 1: BIODEVICES, pages 411-416

ISBN: 978-989-758-398-8; ISSN: 2184-4305

411

Russia - to corrector of the sympathetic nervous

system activity "SYMPATHOCOR-01". During

treatment, the technique of Dynamic Correction of

the Activity of the Sympathetic Nervous System

(DCASNS) was used (Petrenko, Kublanov, &

Retiunskiy, 2015).

Some results of these studies are given in (Darya

D. Egorova, Myakotnykh, Kazakov, & Kublanov,

2014), in which 112 patients participated (40 women,

72 men), with an average age of 45.85 ± 12.41 years.

These patients suffered from stage II – III AH and had

various clinical symptoms of ANS dysfunction.

According to the options for the drugs used the

patients were comparable in the drug

antihypertensive therapy (including angiotensin -

converting enzyme inhibitors, diuretics, slow calcium

channel blockers). Although the effectiveness of

treatment at different periods was ambiguous, which

is quite common in patients with hypertension.

When using the "SYMPATHOCOR-01" device,

the biotropic parameters of spatially distributed

current pulses were selected individually (V. S.

Kublanov, 2008). The treatment course consisted of

10 procedures, and during each procedure:

• 5 minutes of the field exposure in the projection

of the left upper cervical ganglion of the

sympathetic nervous system,

• 5-minute break,

• 5 minutes of the field exposure in the projection

of the right upper cervical ganglion.

Therapeutic sessions were on daily basis. During

treatment, HRV was recorded in dynamics - before

treatment, after the 1st, 5th and 10th procedures of

DCASNS.

The data given in (Darya D. Egorova et al., 2014)

indicate that during the treatment process

normalization of elevated systolic blood pressure was

observed after the 1st session of DCASNS and a

slight decrease in the level of diastolic blood pressure

after the 10th session of DCASNS.

At the same time, the principal component

analysis method was used, which allowed the analysis

of HRV to identify aggregated factor indicators that

are most sensitive to functional changes in the body.

So after the 10th treatment procedure, the parameters

of the clusters of the main components deteriorate,

although, on the other hand, the number of patients in

the cluster with the best indicators increases. A

similar result, according to the authors, may be due to

the fact that when choosing the number of treatment

procedures, the individual dynamics of changes in

HRV parameters were not taken into account.

It is likely that a decrease in the level of systolic

and diastolic blood pressure as a result of the course

of DCASNS should not be considered as the only

indicator of evaluating the effectiveness of this type

of exposure, especially since all patients received

individually selected basic treatment with

antihypertensive drugs. The reaction of regulatory

systems, primarily the ANS, to external therapeutic

effects (for example, DCASNS) may have individual

differences, which must be taken into account when

selecting biotropic parameters of exposure and

determining and duration of treatment.

In the work (Vladimir Kublanov & Dolganov,

2019) presented a new method of forming complexes

diagnostically significant parameters of the HRV for

the diagnosis of AH. A distinctive feature in the

formation of these parameters is the application of

original algorithms based on the paradigm of

evolutionary programming. The formed parameters

complexes allow for express diagnostics with high

values of accuracy, specificity and sensitivity (98.5,

100.0 and 96.0%, respectively) on the training sample

using leave-one-out (LOO) cross-validation

technique, and also has the ability to generalize — the

diagnostic accuracy is 93.0% on the validation

sample.

The objective of this work is to analyze and

discuss possibilities of using complexes of

diagnostically significant HRV parameters to

determine the optimal number of neuro-

electrostimulation procedures.

2 MATERIALS AND METHODS

2.1 Clinical Trials (Part 1)

Original studies (Vladimir Kublanov & Dolganov,

2019) were conducted for two groups of patients:

almost healthy (28 people); and patients suffering

from AH II - III degree before treatment (40 people).

Data were obtained at the Sverdlovsk Regional

Clinical Psychoneurological Hospital for War

Veterans (Yekaterinburg). For registration of

electrocardiogram and HRV signals, the

corresponding recording channel of the

“Eencephalan- -131-03” analyzer was used.

The studies of the HRV signals were carried out

when performing a passive orthostatic test, and

included the state of functional rest (state F), the state

of passive orthostatic load (state O), the state of

aftereffect (state K). HRV signals in each state were

recorded for 5 minutes.

NDNSNT 2020 - Special Session on Non-invasive Diagnosis and Neuro-stimulation in Neurorehabilitation Tasks

412

2.2 Significant Parameters

A complete set of parameters includes statistical

parameters, geometric parameters, spectral

parameters in accordance with the recommendations

of the European Society of Cardiology (Malik, 1996)

and the Commission for Clinical Diagnostic

Instruments and Apparatuses of the Committee on

New Medical Technology of the Ministry of Health

of the Russian Federation, as well as a set of the most

significant non-linear parameters (Sivanantham &

Shenbaga Devi, 2014; Tarvainen, Niskanen,

Lipponen, Ranta-Aho, & Karjalainen, 2014).

In the present study, in addition to generally

accepted parameters, wavelet transform parameters

were used (D.D. Egorova, Kazakov, & Kublanov,

2014). A total of 64 disjoint sets of parameters were

used in three functional states. Since the parameters

in different functional states were considered

separately, the total length of the parameter vector

was 192.

Studies of assessing the accuracy of diagnosing

arterial hypertension using different machine learning

methods and different approaches to the formation of

complexes of diagnostically significant parameters

allowed to establish the following.

The best classification results are achieved using

Genetic programming and Quadratic Discriminant

Analysis. The maximum classification accuracy

(98.5%) using LOO cross-validation was achieved

using complexes of diagnostically significant

parameters consisting of Statistical (Mean, Heart

Rate, Zero-Crossing Rate), Geometric (Stress Index),

Spectral (LF/HF ratio, normalized powers of HRV

spectral components, maximum frequencies of HRV

spectral components), Nonlinear (Entropy), Wavelet

(Entropy of time series HF (t), LF (t), VLF (t))

parameters. In this case, the best solution contains

parameters registered in all functional states (V.

Kublanov, Dolganov, & Gamboa, 2017).

If a quadratic discriminant analysis is used to

predict the class of the test subject (“healthy” or

“patient suffering from hypertension”), the result of

this operation will be the probable class of the test

subject and the probability of belonging to this class.

Since in our case there are only two classes of

subjects, the use of quadratic discriminant analysis

allows us to reduce the multidimensional space of

diagnostically significant parameters to the one-

dimensional space of the metric of the decision rule.

When the classifier is trained, a separating

hyperplane is formed, which “separates” the two

classes of subjects. In the space of the decision rule,

this hyperplane defines the origin. In our case,

positive values of the metric in the space of decision

rules correspond to the class of subjects “healthy”,

negative values of the metric correspond to the class

of subjects “patients suffering from hypertension”.

The table 1 shows combinations of variables

determined using genetic programming that have the

highest classification accuracy (Vladimir Kublanov

& Dolganov, 2019). Combinations are named

according to the number of features in them.

Table 1: Best Parameters Combinations.

Name ACC, % SPE, % SEN, % Features

QDA

-12

98,5 100 96

F HR F EnInterp O kurtosis O IAB

O LF/HF f O LFn wt O SD1/SD2 K IAS

K f

(

LFmax

)

K SDHF K EnHF K EnVLF

QDA

-14-1

98 5 100 96

F SI F EnInterp O kurtosis O ZCR

O LF/HF f O RF O f(LFmax)

O f(VLFmax) O LFn wt K HR

K f(LFmax) K VLFmax K EnHF

EnVLF

QDA

-14-2

98 5 100 96

F SI F EnInterp O HR O kurtosis

O LF/HF f O RF O f(VLFmax) O LFn

wt O EnVLF K f(LFmax) K VLFmax

K HF wt K EnHF K EnVLF

QDA

-14-3

98 5 100 96

F SI F EnInterp O kurtosis O ZCR

O LF/HF f O RF O f(LFmax)

O f(VLFmax) O LFn wt O EnVLF

K f(LFmax) K VLFmax K EnHF

K EnVLF

QDA

-15

98 5 100 96

F SI F EnInterp O kurtosis O ZCR

O LF/HF f O RF O f(LFmax)

O f(VLFmax) O LFn wt O EnVLF

K HR K f(LFmax) K VLFmax K EnHF

K EnVLF

2.3 Clinical Trials (Part 2)

To evaluate the effectiveness of the obtained

combinations of parameters for solving the problem

of diagnosing AH and their ability to generalize, a

study was conducted on data that was not used for

training. This work uses case-study data from 5

patients. In addition to HRV data, blood pressure was

measured using professional tonometer. Data were

obtained at the Sverdlovsk Regional Clinical

Psychoneurological Hospital for War Veterans

(Yekaterinburg).

Diagnostic results were obtained for all 5 patients

before treatment, after 5 procedures, and after 10

procedures for using the biotechnological system of

multichannel neuro-electrostimulation by

"SYMPATHOCOR-01" device.

3 RESULTS

The table 2 provides quadratic discriminant analysis

decision rule metrics, which show the likelihood of

belonging to the particular class for 5 combinations

of parameters that have the best estimates of

Machine Learning Possibilities for Evaluation of Arterial Hypertension Treatment Efﬁciency in Case Study

413

accuracy, sensitivity and specificity in the

classification.

The data is provided for signals recorded

before (0), after 5 and after 10 procedures of neuro-

electrostimulation. The table 3 shows the

measurement data of diastolic and systolic blood

pressure (BPs and BPd) and heart rate (HR) recorded

during the state of functional rest (state F), during the

orthostatic test (state O), and for the stage of

aftereffect (state K).

Table 2: Machine Learning Results.

Patien

QDA-12 QDA-14-1 QDA-14-2 QDA-14-3 QDA-15

P1 - 0 -10,2 -85,0 -54,6 -66,5 -81,6

P1 -5 18,0 9,9 40,8 28,8 32,6

P1 - 10 2,6 -51,5 -29,1 -20,2 -30,8

P2 - 0 -27,5 -15,4 -22,5 -14,0 -11,2

P2 -5 -33,1 -15,0 -21,7 -12,0 -14,8

P2 - 10 -4,9 -48,5 -32,0 -45,4 -45,0

P3 - 0 -56,6 -71,0 -62,6 -70,9 -70,9

P3 -5 -27,1 -8,2 -20,1 -6,9 -5,2

P3 - 10 -24,5 -65,1 -59,2 -68,7 -72,6

P4 - 0 -50,8 -491,8 -527,8 -522,1 -523,9

P4 -5 -16,9 -142,4 -132,0 -150,1 -150,4

P4 - 10 -23,6 -18,5 -21,5 -18,1 -18,4

P5 - 0 -8,4 -33,6 -42,9 -48,9 -48,7

P5 -5 -8,8 -10,0 -7,5 -11,5 -11,6

P5 - 10 -10,4 -15,4 -12,3 -14,0 -14,4

Table 3: Measurements Results.

State F State O State K

Patient

BPs BPd HR BPs BPd HR BPs BPd HR

P1 - 0 112 57 77 95 63 125 126 63 80

P1 -5 114 53 65 100 69 107 118 67 69

P1 - 10 114 60 84 107 69 116 123 62 78

P2 - 0 129 73 53 121 86 74 129 75 51

P2 -5 132 80 55 123 85 68 124 80 68

P2 - 10 107 66 62 111 83 82 113 66 58

P3 - 0 121 73 70 134 90 79 127 77 75

P3 -5 110 68 61 130 86 82 111 67 64

P3 - 10 112 85 60 128 85 84 114 70 74

P4 - 0 131 75 90 131 86 94 136 77 84

P4 -5 145 79 62 142 83 82 146 82 72

P4 - 10 170 93 62 164 105 70 152 86 66

P5 - 0 95 61 56 97 65 65 90 57 58

P5 -5 92 63 60 91 66 67 91 62 59

P5 - 10 92 65 56 82 65 70 86 60 67

It is worth noting that negative estimates of the

likelihood ratio for all patients before the course of

treatment indicates the possibility of generalizing the

combinations obtained.

Patient 1, after 5 procedures, became defined as

healthy (positive value of likelihood ratings).

Changes in state can be associated with normalization

of heart rate in all stages and stabilization of systolic

pressure during an orthostatic test. After 10

procedures, 4 out of 5 combinations evaluate this

patient as belonging to a group of patients, which is

consistent with blood pressure measurements.

For Patient 2, there is a similar behavior of

estimates of combinations of heart rate variability

parameters for records before and after 5 procedures,

which does not contradict the data of instrumental

measurements of blood pressure. An increase

(modulo) the likelihood ratio for 4 out of 5

combinations is consistent with a significant decrease

in systolic pressure. At the same time, a decrease in

the QDA-12 combination score can be associated

with normalization of heart rate. Such a phenomenon

may indicate the sensitivity of various combinations

to various physiological processes.

Patient 3 also has different dynamics of different

combinations. According to the QDA-12

combination, there is a dynamics of assessments

towards relatively healthy patients, after 5 procedures

and maintaining this level after 10 procedures. This is

consistent with a change in the nature of fluctuations

in the parameters of systolic pressure and heart rate

during an orthostatic test. On the other hand, 4

combinations indicate a deterioration after 10

procedures, which is consistent with changes in

diastolic pressure.

Patient 4 - normalization of heart rate, which is

partly consistent with QDA-12, blood pressure varies

greatly.

Patient 5 has underestimated likelihood ratios for

4 of the 5 pre-treatment combinations that are

associated with a variation in heart rate and blood

pressure during an orthostatic test. After 5 and 10

procedures, estimates based on heart rate variability

data change slightly, which corresponds to blood

pressure measurements.

4 DISCUSSION

Comparison of the results obtained using a

discriminant analysis and the measurement results of

blood pressure and heart rate allows to assume a

number of conclusions. In fact, there are two

“families” of parameters combinations. The first

includes a combination of QDA -12, the second

includes the remaining parameters (most of the

parameters in these combinations are repeated).

First of all, different QDA combinations are

consistent with different features of the patients and

different functional states. In particelar, QDA-12

combination is most consistent with the change in

heart rate recorded in the State F. For this

combination, the “most optimal” is a heart rate of

NDNSNT 2020 - Special Session on Non-invasive Diagnosis and Neuro-stimulation in Neurorehabilitation Tasks

414

about 62-65. Moreover, deviations up and down are

equally sensitive for this combination of parameters.

At the same time, other combinations are more

sensitive to changes in heart rate when the patients are

in States O and K.

The use of HRV as a parameter for assessing the

effectiveness of the treatment of arterial hypertension

is certainly justified and has well-known advantages.

The main one is the possibility of individual selection

of treatment depending on the state of autonomic

nervous regulation for a particular patient, and the

possibility of dynamic control of the number of

treatment procedures, the effectiveness and safety of

the therapy.

However, this approach has several limitations.

First, the formation of hypertension is affected by

many complex interacting mechanisms (Bajkó et al.,

2012). Secondly, the functional state and reactivity of

the ANS in hypertension, including the response to

exposure to physical factors, may vary depending on

the degree and stage of hypertension, the presence

and combination of risk factors (smoking,

dyslipidemia , overweight) (Banik, 2014), target

organ damage (brain, kidneys, blood vessels, heart)

(Melillo, Izzo, De, & Pecchia, 2012), the state of the

stiffness properties of the vascular wall (including

their changes in case of damage to target organs of

AH).

It is known that the spectral parameters of HRV

(HF, LF, VLF values) can respond to changes in the

biochemical composition of blood, including the

content of catecholamines, angiotensin II, BDNF, and

products of oxidative stress (Allen, Jennings,

Gianaros, Thayer, & Manuck, 2015; Ferrario et al.,

2015). The psycho-emotional state of the patient has

a huge impact on HRV, including the presence, nature

and severity of anxiety-depressive spectrum

disorders, which are often found in hypertension

(Bajkó et al., 2012; Togo, Kiyono, Struzik, &

Yamamoto, 2006). The dynamic interaction of these

and many other factors, known and not yet fully

studied, can change HRV and the ANS response to

exposure in patients within one general group,

comparable in a number of clinical and biological

parameters (age, gender, degree of increase in blood

pressure, n stage AH). Therefore, HRV cannot always

be used as the only tool for assessing the patient's

condition and receiving feedback in the treatment of

hypertension.

The search for solutions to this problem can be in

the creation of integrated assessment systems (for

example, the previously successfully used

comparison of HRV parameters and indicators of the

blood supply to the brain and the main arteries of the

head (according to ultrasound data), a number of

laboratory test parameters), duration and mode of

exposure to physical factors, identification of key

indicators, determining the effectiveness and duration

of the course of treatment.

5 CONCLUSIONS

The paper describes case-study of the arterial

hypertension treatment using neuro-

electrostimulation device “SYMPATHOCOR-01”

and its efficiency evaluation using methods of

machine learning.

The pathophysiological factors of arterial

hypertension indicate the exceptional role of the

autonomic nervous system violations. Because of that

in previous works was developed the decision support

system based on the most indicative parameters of the

short-term heart rate variability signals (which are

indirect way to assess functioning of the autonomic

nervous system) for the express diagnosing of the

arterial hypertension using methods of machine

learning. In particular, it was shown that original

algorithms of the feature selection based on the

evolutionary programming paradigm allow to obtain

several combinations of heart rate variability

parameters which can be used for the express

diagnosing of arterial hypertension,

In this study was tested applicability of such

approach in task of the treatment efficiency

evaluation. Case study was conducted in which the

neuro-electrostimulation device “SYMPATHOCOR-

01” was used as a method of physiotherapy.

Comparative analysis of the quadratic

discriminant analysis application results and

instrumental measurements highlights concern

regarding using of a single method in such complex

task as a clinical process.

The search for solutions to this problem in the

creation of integrated assessment systems is an

important scientific and practical task.

ACKNOWLEDGEMENTS

The reported study was funded by RFBR according

to the research project № 18-29-02052 and supported

by Act 211 Government of the Russian Federation,

contract № 02.A03.21.0006.

Machine Learning Possibilities for Evaluation of Arterial Hypertension Treatment Efﬁciency in Case Study

415

REFERENCES

Allen, B., Jennings, J. R., Gianaros, P. J., Thayer, J. F., &

Manuck, S. B. (2015). Resting high‐frequency heart

rate variability is related to resting brain perfusion.

Psychophysiology, 52(2), 277–287.

Baevskiy, R. M. (2001). Аnaliz variabelnosti serdechnogo

ritma pri ispolzovanii razlichnykh

ehlektrokardiograficheskikh sistem (metodicheskie

rekomendatsii) [Analysis of heart rate variability using

different electrocardiographic systems (guidelines)].

Vestnik Aritmologii [Herald Arhythmology], (24), 65–

87.

Bajkó, Z., Szekeres, C.-C., Kovács, K. R., Csapó, K.,

Molnár, S., Soltész, P., … Bereczki, D. (2012).

Anxiety, depression and autonomic nervous system

dysfunction in hypertension. Journal of the

Neurological Sciences, 317(1), 112–116.

Banik, S. (2014). Blood pressure and incidence of twelve

cardiovascular diseases: Lifetime risks, healthy life-

years lost and age-specific associations in 1.25 million

people. Journal of Neuroanaesthesiology and Critical

Care, 1(03), 224–225.

da Silva, N. T., Giacon, T. R., Vanderlei, F. M., Vitor, A.

L. R., Pastre, C. M., & Vanderlei, L. C. M. (2014).

Hypertension and autonomic control. American

Journal of Medical Sciences, 2(2), 48–53.

Egorova, Darya D., Myakotnykh, V. S., Kazakov, Y. E., &

Kublanov, V. S. (2014). Application of the major

component for the analysis of heart rate variability in

normal and in hypertension. Journal of Ural Medical

Academic Science, (1), 38–40.

Egorova, D.D., Kazakov, Y. E., & Kublanov, V. S. (2014).

Principal Components Method for Heart Rate

Variability Analysis. Biomedical Engineering, 48(1),

37–41. https://doi.org/10.1007/s10527-014-9412-7

Ferrario, M., Moissl, U., Garzotto, F., Cruz, D. N., Tetta,

C., Signorini, M. G., … Guzzetti, S. (2015). The

forgotten role of central volume in low frequency

oscillations of heart rate variability. PloS One, 10(3),

e0120167.

Kseneva, S. I., Borodulina, E. V., Trifonova, O. Y., & Udut,

V. V. (2016). Cardiac Autonomic Drive during Arterial

Hypertension and Metabolic Disturbances. Bulletin of

Experimental Biology and Medicine, 161(2), 237–240.

Kublanov, V., Dolganov, A., & Gamboa, H. (2017).

Genetic programming application for features selection

in task of arterial hypertension classification.

Proceedings - 2017 International Multi-Conference on

Engineering, Computer and Information Sciences,

SIBIRCON 2017, 561–565. https://doi.org/10.1109/

SIBIRCON.2017.8109954

Kublanov, V. S. (2008). A hardware-software system for

diagnosis and correction of autonomic dysfunctions.

Biomedical Engineering, 42(4), 206–212.

Kublanov, Vladimir, & Dolganov, A. (2019). Development

of a decision support system for neuro-

electrostimulation: Diagnosing disorders of the

cardiovascular system and evaluation of the treatment

efficiency. Applied Soft Computing, 77, 329–343.

https://doi.org/10.1016/j.asoc.2019.01.032

Malik, M. (1996). Heart rate variability: Standards of

measurement, physiological interpretation, and clinical

use. Circulation, 93(5), 1043–1065. Retrieved from

Scopus.

Mancia, G., Fagard, R., Narkiewicz, K., Redon, J.,

Zanchetti, A., Bohm, M., … Zannad, F. (2013). 2013

ESH/ESC Guidelines for the management of arterial

hypertension. European Heart Journal, 34(28), 2159–

2219. https://doi.org/10.1093/eurheartj/eht151

Melillo, P., Izzo, R., De, L., & Pecchia, L. (2012). Heart

rate variability and target organ damage in hypertensive

patients. BMC Cardiovascular Disorders, 12.

https://doi.org/10.1186/1471-2261-12-105

Petrenko, T. S., Kublanov, V. S., & Retiunskiy, K. Y.

(2015). Dynamic Correction of the Activity

Sympathetic Nervous System (Dcasns) to Restore

Cognitive Functions. European Psychiatry, 30, 843.

Sivanantham, A., & Shenbaga Devi, S. (2014). Cardiac

arrhythmia detection using linear and non-linear

features of HRV signal. 2014 International Conference

on Advanced Communication Control and Computing

Technologies (ICACCCT), 795–799. https://doi.org/10.

1109/ICACCCT.2014.7019200

Tarvainen, M. P., Niskanen, J.-P., Lipponen, J. A., Ranta-

Aho, P. O., & Karjalainen, P. A. (2014). Kubios HRV–

heart rate variability analysis software. Computer

Methods and Programs in Biomedicine, 113(1), 210–

220.

Togo, F., Kiyono, K., Struzik, Z. R., & Yamamoto, Y.

(2006). Unique very low-frequency heart rate

variability during deep sleep in humans. IEEE

Transactions on Biomedical Engineering, 53(1), 28–

34. https://doi.org/10.1109/TBME.2005.859783

Ushakov, I. B., Orlov, O. I., Baevskii, R. M., Bersenev, E.

Y., & Chernikova, A. G. (2013). Conception of health:

Space-Earth. Human Physiology, 39(2), 115–118.

https://doi.org/10.1134/S0362119713020175

NDNSNT 2020 - Special Session on Non-invasive Diagnosis and Neuro-stimulation in Neurorehabilitation Tasks

416