Elsevier

Applied Soft Computing

Volume 77, April 2019, Pages 329-343
Applied Soft Computing

Development of a decision support system for neuro-electrostimulation: Diagnosing disorders of the cardiovascular system and evaluation of the treatment efficiency

https://doi.org/10.1016/j.asoc.2019.01.032Get rights and content

Highlights

  • Presents non-intrusive method of arterial hypertension diagnostics.

  • Heart rate variability signals registered during tilt-test were used.

  • Variety of feature selection and machine learning methods were tested.

  • Results suggest possibility of diagnostics and evaluation of treatment efficiency.

Abstract

The study describes a preliminary stage of the decision support system development for physician performing neuro-electrostimulation of neck neural formations for patients suffering from cardiovascular system disorders. The arterial hypertension was used as the clinical model of the disorders. The study consisted of two steps: diagnosing of the arterial hypertension and an evaluation of the treatment efficiency during the neuro-electrostimulation application. For the diagnosing part, a clinical study was conducted involving heart rate variability signals recorded while performing tilt-test functional load. Heart rate variability signal is an indirect mean of accessing autonomic nervous system functioning. Disturbances of the autonomic nervous system are essential in pathology of arterial hypertension. Performance of different machine learning techniques and feature selection strategies in task of binary classification (healthy volunteers and patients suffering from arterial hypertension) were compared. The genetic programming feature selection and quadratic discriminant analysis classifier reached the highest classification accuracy. Best feature combinations were used to evaluate treatment efficiency. Predictions based on the selected heart rate variability features have a high level of agreement with the arterial pressure dynamics. The results indicate the potential of the proposed decision support system.

Introduction

All living beings are constantly adapting to changing factors of the environment for maintaining conditions essential for normal life. For the role of constant internal environment maintenance (homeostasis) as an evolutionary product cardiovascular system has arisen. This is a complex network, dedicated to transport of various liquids [1]. Cardiovascular system disorders can remain unnoticed for a rather long time. However, abruptly they can cause acute impairments [2]. As noted in the World Health Organization reports, the heart failure and stroke, regrettably, remain the leading causes of death in the world [3]. Therefore, the task of the early pre-clinical express diagnosing is relevant.

Among the variety of reasons, causing disorders of the cardiovascular system normal functioning, one can highlight etiological factors. These factors mainly influence the vascular wall, changing its structure and causing disorder of the vascular tone. Moreover there are factors that initiate heart pathologies. In particular ones, causing inflammatory and dystrophic processes [4]. These factors might cause arterial hypertension. As noted in World Health Organization [5] arterial hypertension is among the most frequent cardio-vascular pathologies and occurs in around 15%–20% of the elderly population. The hypertension is considered to be among the most prominent factors of the heart failure, stroke and coronary failure. Therefore, arterial hypertension is a suitable clinical model for cardiovascular system disorders.

The nature of the arterial hypertension is multifactorial. The hypertension arises as the disorder of the vascular wall tone. The vascular tone’s most important feature is the arterial pressure. Which is in turn controlled by the regulatory mechanisms within the autonomic nervous system (ANS) [6]. As pointed out in guidelines for management of arterial hypertension [7], the arterial pressure is a crucial factor, but not the only factor, which defines hypertension severity, diagnosis and treatment course. Among the challenges in the diagnosing of the cardiovascular system disorders are situational changes, like stress. Thus it is appropriate to develop an objective decision support system for minimizing these factors [8].

A detailed analysis of current research in the field of artificial intelligence application for prediction and management of hypertension is done in [9]. It is worth noting that most of the researches in this area are devoted to predicting the risk of hypertension. In studies as a rule, clinical research data are used (parameters of the basic physician examination, blood routine items, urine routine items, gene expression data, etc.), ambulatory blood pressure monitoring, and data on the environment and lifestyle (working conditions, diet, hereditary diseases). Moreover, in this comparative review our previous study [10] was noted as the only one in which short-term signals of heart rate variability (HRV) were used as input variables. Research groups around the globe look for different means of arterial hypertension diagnosing. For example group of Pytel [11] has used data of anthropomorphic features for evaluation of the hypertension degree by means of the neural networks. Alternatively, in [12] the possibilities of support vector machine was shown in task of the essential hypertension classification using environmental and genetic factors. Even though, those studies have reached good level of accuracy there are some shortcomings. The most significant one is lack of possibility to quantify dynamics of the physiological changes during the treatment.

Often, in accordance with recommendations of the medical societies, arterial hypertension treatment consists of several medications. These may lead to multiple side effect [7]. The alternative to the pharmacological approach is application of the physiotherapy devices dedicated to improvement of the cardiovascular system. The SYMPATHOCOR-01 neuro-electrostimulator is one of such devices. The device provides correction of the sympathetic department of the ANS. Through the sympathetic department the device has constricted control of the vascular tone [13]. Thereby it is appropriate to apply in the development of the decision support system for treatment of arterial hypertension such methods that allow monitoring current state of the ANS.

The HRV is one of the indirect means of the ANS monitoring. The heart rate varies from beat to beat. This is caused by constant adaptation processes, which are launched by the ANS to keep balance of the cardiovascular system. The HRV reflects the functioning of the cardiovascular system and regulatory mechanisms of the organism, as well as the ratio between sympathetic and parasympathetic departments of the ANS. Changes of the HRV features can be preemptive indicators of the health disorders [14].

However, some works state that the prognostic possibilities of the HRV signals are limited. Alternatively, combination of standard features in time–frequency domain and non-linear features can improve diagnostic possibility for better identification of patients at risks [15]. It is advisable to use as much features from the signal as possible. At the same time, one should also consider proper feature selection to avoid using redundant data which can make diagnosing task harder [16].

In order to minimize situational deviations influence on the diagnostics results it is appropriate to consider functional loads. In this case one can get information that is fully reflecting pathophysiological state of the patient by choosing particular load that activates response of the pathological function. As an example of the functional load one can consider the tilt-test [17]. The tilt-test is an experimental way to test organism reaction on the crossover from the horizontal position to the vertical one (head up). Usually the rotating table is used for this load. Among the benefits of the rotating table application is increased sensitivity, improved reproducibility of conditions and results, safety and possibility to control the angle of the rotation as well as its speed. Finally it reduces the noise of the biomedical signals registration, as it prevents active body movement [18]. The organism reaction to the tilt-test is well studied for both, healthy persons and in case of several pathologies. As studied are the responses of the cardiovascular system. Therefore, one can compare changes of the HRV during the tilt-test with the well-known physiological reactions of the cardiovascular system.

In view of above the goal of the work is to develop decision support system for neuro-electrostimulation of neck neural formations for patients suffering from arterial hypertension. The following tasks are considered:

  • development of the methodologies for application of the machine learning techniques to heart rate variability signals for diagnosing of the arterial hypertension by

    • testing classification efficacy for signals recorded during functional loads,

    • comparing different feature selection strategies

    • comparing classification efficacy of different machine learning techniques;

  • preliminary analysis of the neuro-electrostimulation treatment efficiency evaluation.

The generalized workflow of the study is presented on Fig. 1.

The structure paper is as follow: Section 2 presents materials and methods, in particular clinical study description (Section 2.1), HRV features depiction (Section 2.2), studied Machine learning techniques (Section 2.3) feature selection strategies (Section 2.4). In Section 3 we describe results, in particular evaluation of machine learning techniques and feature selection strategies application for diagnosing of arterial hypertension (Section 3.1), preliminary analysis of possibilities for results application in evaluation of treatment efficiency (Section 3.2). In Section 4 we discuss results of the study and propose algorithm for application in physician decision support. Section 5 is the Conclusion.

Section snippets

Clinical study description

The pilot clinical study was approved by the local Ethics Committee in Ural State Medical University, Yekaterinburg, Russian Federation (protocol No. 8 from 15 October 2015). Overall, 68 people have participated in the study: 28 healthy people and 40 patients. All the patients were diagnosed with the II/III degree of arterial hypertension. All participants were volunteers: prior to the study, they were given detailed explanation of the study paradigm. After that, they had signed the written

Arterial hypertension diagnosing

In the study the classification accuracy of two groups – healthy and patients, suffering from arterial hypertension – was evaluated for each of the machine learning technique. First of all, basic classification with all features was performed. Table 5 presents the results of such classification. Data in Table 5 shows that basic classification achieves non-optimal results for all feature vectors. In case when all features were used the maximal accuracy was achieved by k-NN classifiers when F-O

Discussion

Materials and methods used in current study vary from those used in state-of-art. In works that use limited number of features exhaustive search was sufficient [46]. However, in this study such approach would not be appropriate as great number of features leads to enormous number of possible combinations. Therefore variety of feature selection and dimensionality reduction techniques were tested.

Application of semi-optimal search in non-correlated space allowed reaching relatively high results

Conclusions

Cardiovascular diseases are the leading cause of death, despite the availability of effective and inexpensive approaches to treatment. Therefore, the development of new methods for diagnosing and predicting the development of these diseases is undoubtedly an urgent task. The present paper described results of the study which can be used for combined treatment of patients, suffering from the cardiovascular system disorders. As the clinical model of disorders the arterial hypertension was used.

Acknowledgments

The reported study was funded by RFBR, Russia according to the research project No. 18-29-02052.

The authors thank Doctor Yan Kazakov, Ural State Medical University, for consultations and discussion of the results.

References (49)

  • MohrmanD.E. et al.

    Cardiovascular physiology

  • MendisS. et al.

    Global Atlas on Cardiovascular Disease Prevention and Control

    (2011)
  • MohrJ.P. et al.

    Stroke E-Book: Pathophysiology, Diagnosis, and Management

    (2011)
  • A Global Brief on Hypertension: Silent Killer, Global Public Health Crisis: World Health Day 2013

    (2013)
  • KsenevaS.I. et al.

    Cardiac autonomic drive during arterial hypertension and metabolic disturbances

    Bull. Exp. Biol. Med.

    (2016)
  • ManciaG. et al.

    ESH/ESC guidelines for the management of arterial hypertension

    Eur. Heart J.

    (2013)
  • KrittanawongC. et al.

    Future direction for using artificial intelligence to predict and manage hypertension

    Curr. Hypertens Rep.

    (2018)
  • DolganovA.Y. et al.

    Comparison of machine learning methods for the arterial hypertension diagnostics

    App.l Bionics Biomech.

    (2017)
  • PytelK. et al.

    Anthropometric predictors and artificial neural networks in the diagnosis of hypertension

    (2015)
  • PeiZ. et al.

    Risk-predicting model for incident of essential hypertension based on environmental and genetic factors with support vector machine

    Interdiscip. Sci. Comput. Life Sci.

    (2018)
  • KublanovV.S.

    A hardware-software system for diagnosis and correction of autonomic dysfunctions

    Biomed. Eng.

    (2008)
  • AhmadS. et al.

    Clinical review: A review and analysis of heart rate variability and the diagnosis and prognosis of infection

    Crit. Care

    (2009)
  • Ducla-SoaresJ.L. et al.

    Wavelet analysis of autonomic outflow of normal subjects on head-up tilt, cold pressor test, Valsalva manoeuvre and deep breathing

    Exp. Physiol.

    (2007)
  • TurkU. et al.

    Prediction of head-up tilt test result: Is it possible

    Pacing Clin. Electrophysiol.

    (2010)
  • Cited by (15)

    View all citing articles on Scopus

    This paper is an extended, improved version of the paper ‘Towards a Decision Support System for Disorders of the Cardiovascular System - Diagnosing and Evaluation of the Treatment Efficiency’ presented at AI4Health 2018 workshop and published in: BIOSTEC 2018, Proceedings of the 11th International Joint Conference on Biomedical Engineering Systems and Technologies, Volume 5: HEALTHINF, Funchal, Madeira, Portugal, 19-21 January, 2018, pp. 727-733, ISBN: 978-989-758-281-3, INSTICC, 2018.

    View full text