Cardiovascular changes and arrhythmia risk stratification in young and elite athletes
- (0) Download
https://www.eduzhai.net American Journal of Biomedical Engineering 2014, 4(3): 60-67 DOI: 10.5923/j.ajbe.20140403.02 Cardiovascular Modifications and Stratification of the Arrhythmic Risk in Young and Master Athletes R. Pizzi1,*, S. Siccardi1, C. Pedrinazzi2, O. Durin2, G. Inama2 1Department of Computer Science, Universitàdegli Studi di Milano, Crema Campus, Crema 26013, Italy 2Department of Cardiology, Ospedale Maggiore di Crema, Crema 26013, Italy Abstract This paper aims to evaluate the cardiovascular response to exercise in young and master athletes. The study involves comparison with control groups of untrained subjects of the same age. The paper describes in particular the processing methods and results concerning the ECG signals collected during exercise test. Methods include PNN calculation, Multiscale Entropy analysis (MSE), and a comparison between clustering and an Artificial Neural Network analysis performed by means of chaotic attractors. The analyses carried out lead to a good stratification of the subjects, especially in terms of the MSE1 variable. Future developments are underway supported by many other diagnostic tests administered to the same subjects. Keywords Electrocardiography, Clustering methods, Artificial Neural Networks, Data Mining 1. Introduction The project aims to evaluate the cardiovascular response to exercise in young and master athletes. The study involves comparison with control groups of untrained subjects of the same age. The complete surveys provided during the study have been: 12-lead resting ECG, cardiopulmonary exercise test, venipuncture with doses of various parameters, echocardiogram, ambulatory ECG, T-wave alternans (TWA). This paper describes in particular the processing methods and results concerning the ECG signals collected during exercise test. The presence of cardiac remodeling in elite athletes is a concept known ever since the last decade of XIX century, when the percussion techniques of the cardiac area allowed to identify an increase in the volume of the heart in subjects who practiced competitive sport. Later, the introduction into clinical practice of echocardiographic techniques has allowed us to better understand cardiac physiology of the athletes and the impact that different modes of sport and exercise training on cardiac structure and function, allowing to characterize in detail the framework of the "athlete heart ". In addition, modern methods of non-invasive and invasive electrophysiology contribute to deepen the possible correlation between the type and intensity of athletic * Corresponding author: email@example.com (R. Pizzi) Published online at https://www.eduzhai.net Copyright © 2014 Scientific & Academic Publishing. All Rights Reserved activity and cardiac arrhythmias [2, 3, 4, 5, 6, 7]. Most of the sudden deaths in young athletes, indeed, appear to be due to asymptomatic heart diseases, that are accelerated and unmasked by a particularly intense physical activity. In particular, hypertrophic cardiomyopathy is responsible for 30% of fatal cases in the United States while about 25% in Italy is caused by arrhythmogenic right ventricular cardiomyopathy. A recent clinical trial of endurance athletes during a 17 years follow-up showed that vigorous exercise may itself affect a structural remodelling, in particular of the right ventricle, that can sometimes predispose to development of potentially fatal ventricular arrhythmias [8, 9, 10, 11, 12, 13, 14]. In particular, the physical exercise is associated with hemodynamic changes with alteration of the heart loading conditions. Endurance sports are aimed at the production of movement (e.g. running) with limited development of muscle strength. In this case from the hemodynamic point of view an increase of heart rate and stroke volume is observed which are the two determinants of cardiac output. Peripheral vascular resistances are reduced with only a slight or moderate increase in blood pressure. In this type of sport the heart load is primarily a volume load, that may lead to a dilatation of the left ventricle with a proportional increase in wall thickness (eccentric left ventricular hypertrophy). On the contrary in power sports (e.g. weightlifting, rowing etc) there is strength development with limited movement. In this case, the hemodynamic consequences include only a slight increase in cardiac output favored by the increase in frequency and a more pronounced increase in blood pressure American Journal of Biomedical Engineering 2014, 4(3): 60-67 61 caused by increased peripheral vascular resistance. These phenomena lead to an increase of pressure load on the heart and in the long term may favor the development of concentric left ventricular hypertrophy with increase in wall thickness and maintenance of normal left ventricular cavity size. While there are numerous works and studies that evaluate the impact of sport activity on the cardiovascular system in youth, in the literature there are very few data on cardiovascular adaptation to sport activity in master athletes. In particular, there are no prospective and controlled researches that compare in a comprehensive way the two populations of young athletes (<40 years) and "Master" (> 40 years old) in terms of the cardiovascular adaptation to training. An adequate knowledge of the physiological mechanisms of adaptation to exercise in a subject with age> 40 also assumes an important clinical significance particularly for the development of cardiac rehabilitation programs after myocardial infarction and coronary revascularization and for the prescription of appropriate physical training in the post-rehabilitation phase.  In this regard it should be noted that the type of exercise (combined resistance or interval training), most appropriate intensity and frequency of training sessions in subjects with heart disease are still subject of study and controversy. In addition to young competitive athletes, several adults and elderly patients, in some cases affected by cardiovascular diseases practice leisure physical activity, only in order to achieve or maintain a state of health and well-being. Favorable effects of exercise are known, not only in cardiac metabolism, body weight and blood pressure but also on improving functional capacity, the reduction of cardiovascular events and the increase of endothelial progenitor cells that appear to have a role important in angiogenesis, the inhibition of intimal proliferation and vascular repair. And for these reasons, physical activity and sport should be part of clinical management of patients with heart disease [15, 16, 17, 18]. Physical activity must be understood as a prescribed drug, specifying in each case indication, route of administration and doses, not to mention, as with all medicines, problems of overdose and side effects. With the study conducted in collaboration between the Division of Cardiology, Ospedale Maggiore of Crema and the Department of Computer Science, University of Milan, we mean to evaluate the cardiovascular response to exercise in young and master athletes, even by comparison with control groups of untrained subjects of the same age. We also meant to identify the presence in athletes, of possible markers of increased arrhythmic risk. 2. Materials and Methods We considered four groups of subjects: 1) Group A: athletes practicing endurance sports (cycling or running) with age <40 years 2) Group B: "master" athletes practicing endurance sports (cycling or running) and> 40 years old 3) Group C: healthy subjects aged <40 years not practicing sports 4) Group D: healthy subjects aged> 40 years not practicing sports. This report describes the processed data concerning the ECG signals collected during exercise test. Figure 1. The graph represents the typical stress ECG behavior for six subjects randomly chosen, where heartbeat increases regularly from the values at rest (between 70 and 100 beats per second). After reaching a maximum, it decreases again in the recovery phase 62 R. Pizzi et al.: Cardiovascular Modifications and Stratification of the Arrhythmic Risk in Young and Master Athletes The sample under study was composed of 20 subjects in group A and 20 in group B, 10 in group C and 11 in group D. Data of the exercise test were also analyzed through an ad hoc software environment and the use of non-linear analysis. At the end of a first data screening we selected usable data on the following subjects: 1) a group of 19 athletes, "master", i.e. over the age of 40, consisting of 6 female athletes and 13 male 2) a group of 18 young athletes, 5 females and 13 males 3) a control group of 7 sedentary, age greater than 40 years, 2 female and 5 male 4) a control group of 8 young sedentary, 3 female and 5 male The analyzes focused on heart rate variability (HRV), whose importance has been recognized in several studies (eg. ). The first step was to derive the instantaneous beat signals by eliminating artifacts. As an example we report in Fig. 1 the graph describing the HRV of some patients. Then "RR intervals" were calculated, i.e. the duration of the pulse in seconds for all the tracks, with the sole exception of the artifact portions. Using these data, the following analyses were carried out: 2.1. PNN Calculation The method takes into consideration the increase in milliseconds from an RR interval to the next and calculates the distribution. Furthermore () it considers individual indicators, such as PNN50 (the probability that the difference between two successive intervals is at least 50 MSEc.) and PNN20 (the probability that the difference between two successive intervals is at least 20 MSEc). According to the literature , young healthy subjects had significantly higher values of the PNN compared to subjects older or suffering from heart. After several tests, we focused on PNN20 that gave better discrimination among subjects. 2.2. Multiscale Entropy Analysis (MSE) The method quantifies the complexity of the data series, taking into account the fact that the physiology dynamics occurs according to multiple time scales (see ). MSEX, i.e. the Multi Scale entropy with scale factor x, is calculated from the data set, constructing the series of averages of the elements taken in groups of x and calculating the sample entropy. So MSE1 = usual sample entropy, MSE5 = sample entropy of the averages of groups of 5 data, MSE20 of groups of 20 data. Literature () reports that the method was able to separate healthy patients from patients suffering from Congestive Heart Failure (CHF) and Atrial Fibrillation (AF), and young patients by elderly. 2.3. Clustering Using anagraphic and behavioural (sedentary / sports) data and the new processed data, we proceeded to the clustering procedure. We used hierarchical clustering with Ward's method, which minimizes the variance internal to the cluster. Initially, each object is assigned to a cluster, the algorithm proceeds iteratively by bringing together the most similar clusters at each step and proceeding until a single cluster remains. The Ward's method was chosen after some testing with alternative methods, because it gave the most satisfactory clusters. For a discussion of the method see [22, 23, 24, 25]. 2.4. Analysis by Artificial Neural Network Another type of classification was carried out using a custom self-organizing neural network. Artificial Neural Networks and in particular the Self Organizing Maps (SOM) are nonlinear models suitable to classify complex patterns [26, 27]. In particular we used a model of SOM-like network developed by our group that highlights chaotic attractors within the sequence of winning neurons. A fruitful use of this model in cardiology has already been proposed by some of the authors . 2.5. Software The calculation of the instantaneous heartbeat, of RR intervals, PNN, multiscale entropy and multifractal analysis were performed using the Physionet library . The analysis of the clusters was performed with R . The neural network and dynamic analysis system have been developed by the group. 3. Results The first clusters were performed with all available variables, namely: 1) athlete yes / no 2) age 3) PNN20 4) MSE1 5) MSE5 6) MSE20 7) Gender 8) Edited data for the presence of artifacts yes / no The analysis was subsequently refined to isolate the variables able to cluster the data in a satisfactory manner. A satisfactory result was obtained considering the variables: 1) athlete yes / no 2) age 3) MSE1 The clustering is represented in Fig. 2, in which the patients belonging to the same group were assigned the same color. American Journal of Biomedical Engineering 2014, 4(3): 60-67 63 Figure 2. Clustering using athlete yes/no, age, MSE1. Here: Sedovm and sedovf = Sedentary over 40 (males and females), Sedunm and sedunf = Sedentary under 40 (boys and girls), Atovm and atovf = Athletes over 40 (males and females), Atunm and atunf = Athletes under 40 (boys and girls) As seen from Fig. 2, 6 clusters are highlighted: 1) The sedentary under 40, characterized by low MSE1 (maximum = 0.182 min = 0.081, mean = 0.15 standard deviation = 0.034). The PNN20, as for most of the clusters, is dispersed: minimum = 0.9, maximum = 19.8, average 11.3, standard deviation 7.66. 2) The sedentary over 40, with highest MSE1 (maximum = 0.381 min = 0.122, mean = 0.222, standard deviation = 0,09). We note that also the PNN20 is greater than in the previous cluster, even if the dispersion is remarkable: PNN20 minimum = 1.7, maximum = 52.1, average 16.3, standard deviation 16.6. In this group also appears a sedentary under 40, which has MSE1 = 0.307, sharply higher than that of the previous group. 3) Athletes over 40, with MSE1 low (maximum = 0.264 min = 0.001 Mean = 0.127, standard deviation = 0.071). The other indicators tend to be low: PNN20 minimum = 2.6, maximum = 23.9 average 11 standard deviation 6.8. An under 40 athlete is in this group and has MSE1 =0,192 that is intermediate between those of the two following cluster, in which there are other athletes under 40. 4) A cluster consists of athletes with high MSE1 (maximum = 0.568 min = 0.297, mean = 0.399, standard deviation = 0,11). It is the group with the highest MSE1 sample. The other indicators are high: PNN20 minimum = 9.5, maximum = 52.4 average 24.4 standard deviation 14. Three of the members are over 40 and four under 40. From the available data no other characteristics are detected. 5) A cluster of athletes under 40 with low MSE1 (maximum = 0.162, minimum = 0.069, mean = 0.115, standard deviation = 0.035). The other indicators are low, in particular in the PNN20 that this group is not dispersed as in the other (minimum = 2.3, maximum = 8.7, average 2.5, standard deviation 5.97). Also note that this group has an average age of 35 years old, so it is older than the next cluster, which contains the other group of athletes under 40. 6) A cluster of young athletes very less than 40 years (mean age 19 years), with MSE1 higher than the previous cluster (maximum = 0.249, minimum = 0.095, mean = 0.137, standard deviation = 0.058). PNN20 values: minimum = 4, maximum = 26.4, average 8.5, standard deviation 11.38. The clusters discriminate on natural way based on the presence or not of a sporting activity, whereas it becomes immediately evident that sex and other variables are not discriminating, with the exception of MSE1. The subjects aged less than 40 years have low MSE1, whether athletes (cluster 5) or non-athletes (cluster 1). Nevertheless athletes also have low PNN20 (Range 2.3 - 8.7 mean 5.97 2.53 standard deviation), which discriminates against them by non-athletes (range .9 - 19.8 average 11.03 standard deviation 7.66) according to this parameter. We note that in general PNN20 is a variable general enough dispersed in all the clusters, however, in the cluster 5 standard deviation is relatively low. Those athletes aged very less than 40 years (cluster 6) have instead MSE1 apparently dispersed, but it becomes low by eliminating one subject with MSE = 0.249 (which has no other special features evident, at least in the data examined so far). This seems to indicate that MSE1 changes after the adult 64 R. Pizzi et al.: Cardiovascular Modifications and Stratification of the Arrhythmic Risk in Young and Master Athletes age, independently from the fact that the subject is an athlete or not. In young athletes instead MSE1 assumes variable values, indicating the hypothesis that in time the full physical maturity causes a cardiovascular modification that lowers MSE1 independently from the physical activity. Conversely, it appears that MSE1 rises in non-athletes over the age of 40, while the athletes of the same age group have a MSE1 almost always low, as if with age physical activity tended to preserve a low MSE1: the comparison between the non-athletes over 40 (cluster 2) and athletes (cluster 3) shows MSE1 lower in athletes. Even PNN20 is smaller, however the dispersion is very high, especially in non-athletes. Interestingly, there exists a separate group of athletes (Cluster 4) with very high MSE1 with other indicators, in turn, with high values independently from the age. The data are quite dispersed: PNN20 minimum 9.5, maximum 52.4 average 24.4 standard deviation 14, MSE1 maximum = 0,568 Minimum = 0.297, mean = 0.399, standard deviation = 0.11. Another feature of the cluster is that 6 of the 7 components have around age 40 years (39 to 45), the other component is 26 years old and the values of PNN20 and MSE1 are the maximum of the cluster. Eliminating this component reduces the dispersion of the variables, especially the MSE1: PNN20 range 9.5-26 average 19.79 standard deviation 7.4; MSE1 range 0.3-0.53 average 0.09, standard deviation 0.09. It would be interesting to follow these subjects and study their follow-up with respect to that of the athletes with low MSE1. The most significant results with the Artificial Neural Network system have been obtained by employing a configuration of 500 input neurons, 10 mapping units, learning rate 0.001, delta = 0.1 and are summarized in the following Tables I-III. Two kinds of chaotic attractors are found: "fixed point", consisting mainly of a single point and prefixed by F in the tables, and "cyclic", consisting of at least two points and prefixed by C in the tables. For instance, F3 is the attractor consisting mainly of the repetition of point 3 only, while C(0-1) is the attractor consisting mainly of the repetition of points 0 and 1. We can also draw several objective considerations. First we note again that sex is not a discriminant variable: there is no evidence of attractors consisting mainly of males or females. Additionally, there is no classification within the group of non-athletes related to the age, while athletes have cardiovascular characteristics differentiated depending on the age. Table 1. Athletes under 40 Attractor C(0-1) F3 F4 C(5-9) F6 F8 N. patients 3 4 1 6 2 2 Average age 24 31 19 36 20 40 PNN20 average 7.033 25.7 26.4 8.2 6.58 20.42 Mse1 average 0.115 0.26 0.249 0.159 0.137 0.295 Mse5 average 0.20 0.35 0.473 0.27 0.19 0.486 Mse20 average 0.5 0.35 0.4 0.6 0.45 0.89 Attractor C(0-7) C(1-4) F2 F3 C(6-9) N. patients 8 5 2 1 3 Table 2. Athletes over 40 Average age 48 48 51 56 44 PNN20 average 12.93 11.43 14.72 2.62 8,135 Mse1 average 0.216 0.112 0.094 0.158 0.165 Mse5 average 0.276 0.138 0.256 0.182 0.196 Mse20 average 0.69 0.31 0.53 0.46 0.53 Attractor C(0–others) C(1-8) C(6-7) F9 N. patients 7 5 2 1 Table 3. Sedentary Av. PNN20 age average 40 19.26 34 11.65 39 5.08 43 4.18 Mse1 average 0.219 0.164 0.16 0.15 Mse5 average 0.31 0.283 0.319 0.17 Mse20 average 0.5 0.58 0.64 0.26 American Journal of Biomedical Engineering 2014, 4(3): 60-67 65 Moreover it is noted that the comparison with the procedures of clustering shows a correspondence between an attractor of the ANN (attractor 5) and the cluster 5 (athletes younger than 40 years old, low MSE1): attractor 5 includes 5 of the 7 subjects of the cluster 5, plus a subject that is not part of the cluster. The variables values for the attractor are the following: MSE1 08 to 0.37, average 0.159, standard deviation 0.1, PNN20 range 2.27-24.23 average 8.2. We note that MSE1and PNN20 have values quite close to those of cluster 5, however, the presence of the outlying subject in the attractor increases the standard deviation. Removing this subject, which has the highest values in all the variables, we get MSE1 range 0.08-0.162 average 0.117, standard deviation 0.03, PNN20 range 2.27-7.68, average 5, standard deviation 2.33, much better aligned to the values of the cluster 5. Fig. 3 shows a graph of the attractor C(1-8) and Fig. 4 shows a graph of the attractor C(0-1). For the cluster 2 we have: MSE1 range 0.122 to 0.381, average 0.22, standard deviation 0.09; PNN20 range 1.7 to 52.1, average 16.3, standard deviation 16.6. Finally, many attractors of non-athletes are composed mainly from elements belonging to the cluster of non-athletes under the age of 40 years. For example, four of the five subjects of the attractor 1-8 are part of the cluster 1. Fig. 3 shows a graph of this group. The results of the Artificial Neural Network (ANN) procedure show that athletes and non-athletes are clearly discriminated on the basis of the already mentioned variables, as the tables show that these groups have no common attractors. The above considerations show that the adopted data analysis procedures are mutually congruent in discriminating the studied subjects, confirming the validity of the obtained results. 4. Conclusions Figure 3. A graph of the attractor C(1-8) Figure 4. A graph of the attractor C(0-1) In addition there is also correspondence between the attractor 0 and the cluster 2 (non-athletes with low MSE1), which have 5 elements in common of 7 and 8 elements respectively. Variables for the attractor have the following values: MSE1 range 0.08 to 0.38, average 0.22 standard deviation 0.1; PNN20 range 1.7-52.1, average 19.26, standard deviation 16.45. Our study has highlighted some interesting aspects about the adaptation of the cardiovascular system to physical exercise and the peculiarities of such adaptation respectively in subjects aged less than 40 years and in master athletes, issue about which literature is quite scarce. Through the non-linear analysis of the data of exercise test we compared two different classification methods in order to crossvalidate the results. Actually we found some concordances between the ANN and clusters classification. For example, the attractor 5 of athletes under 40 consists almost entirely by members of the fifth cluster (athletes under 40 with MSE1 low), the attractor 0 consists of sedentary members of the cluster 2 (under 40 with sedentary MSE1 bass) and the attractor 1-8 of members of the cluster 1 (sedentary under 40). It is also interesting to note that the Artificial Neural Network completely separates sedentary subjects from athletes, as they show no common attractors. We can conclude that the use of non-linear analysis on data from the stress ECG helped to identify a significant division into clusters among the various groups of subjects according to the degree of physical activity, confirming the significant role in determining the functional parameters and cardiovascular diseases. In this analysis we have highlighted both the existence of sensitive variables (athlete / no, age, MSE1) and the existence of significant groups of patients around to these variables. The main conclusions that can be drawn are as follows: 1) The subjects aged less than 40 years have low MSE1, whether athletes or non-athletes. However, athletes have PNN20 low. 2) This would suggest that MSE1 modifies itself above the adult age independently from the physical activity. 3) In young athletes, instead, MSE1 assumes variable values, indicating the hypothesis that over time the full 66 R. Pizzi et al.: Cardiovascular Modifications and Stratification of the Arrhythmic Risk in Young and Master Athletes physical maturity results in a cardiovascular modification that lowers MSE1 regardless of physical activity. 4) Conversely, it appears that MSE1 rises in non-athletes over the age of 40, while the athletes of the same age group have MSE1 almost always low, as if with age, physical activity tended to preserve a low MSE1. 5) The sex is never a discriminating variable. 6) Sedentary subjects are sharply discriminated from athletes by the ANN. 7) It is also worth observing that the ANN classification stresses the lack of age classification within groups of non-athletes, while athletes result to have cardiovascular characteristics differentiated by age. 4.1. Future Developments This preliminary analysis suggests that individuals examined exhibit a distinct cardiovascular status for sedentary lifestyle, age and self-similarity of the signal. Thus we are now interested in evaluating whether and to what extent, within the clusters identified, other variables may indicate subgroups of individuals recruited that exhibit the characteristics that may represent common cardiovascular prognostic indicators. For both the group of athletes and for the control group we collected numerous blood chemistry data, as well as data relating to echocardiography, ambulatory ECG and cardiopulmonary evaluation .Within these data, we have identified 30 key variables, the analysis of which, compared to the clustering already carried out, could provide interesting information in perspective. These results should then be detailed on samples of greater size and with ad hoc analysis in order to clarify the clinical and pathophysiological significance. The subjects participating in the study underwent, as well as exercise test, also to the following diagnostic tests: 1) History and physical examination 2) Venipuncture with a dose of blood count, urea, creatinine, total cholesterol, HDL cholesterol, triglycerides, blood glucose, BNP, CPK, LDH, AST, ALT, sodium and potassium. 3) 12-lead resting ECG 4) Trans-thoracic echocardiogram, with assessment of cardiac morphology and function. and Doppler examination. 5) Evaluation of the maximum O2 consumption (VO2 max), O2 consumption at the anaerobic threshold (VO2AT) and VE/VCO2 slope (indicator of ventilatory response to exercise) during cardiopulmonary exercise testing. 6) 24-hour ambulatory ECG recording, with quantification of ventricular and supraventricular ectopies and search for possible presence of pathological bradyarrhythmias or conduction defects. 7) T-wave alternans. The analysis of the data collected will be presented in a subsequent work. The analysis shall focus in particular the data on the consumption of O2 and laboratory analysis, the study of correlations between the data of this new group and examination of the correlations between these variables and other data calculated so far (entropy, PNN) and used for the clustering presented in this work. In the following a new classification will be carried out, including a comparison of the new results with those of the present study. REFERENCES  H. L. Kennedy, J. A. Whitlock, 1985, Long-Term Follow-up of Asymptomatic Healthy Subjects with Frequent and Complex Ventricular Ectopy,” N Engl J Med, 321,(4), 193-197.  H. Bjornstad, L. Storstein, 1994, Ambulatory electrocardiographic findings in top athletes, athletic students and control subject, Cardiology, 84(1), 42-50.  O. Durin, C. Pedrinazzi, G. Donato, R. Pizzi, G. Inama, 2008, Usefulness of nonlinear analysis of ECG signals for prediction of inducibility of sustained ventricular tachycardia by programmed ventricular stimulation in patients with complex spontaneous ventricular arrhythmias, Annals of Noninvasive Electrocardiology, 13(3), 219-227.  R. Pizzi, G. Inama, O. Durin, C. Pedrinazzi, 2007, Non-invasive assessment of risk for severe tachyarrhythmias by means of non-linear analysis techniques, Chaos and Complexity Letters, 3(3), 229-250,  G. Inama, C. Pedrinazzi, O. Durin, M. Nanetti, R. Pizzi, 2008, Microvolt t-wave alternans for risk stratification in athletes with ventricular arrhythmias: correlation with programmed ventricular stimulation, Annals of Noninvasive Electrocardiology, 13(1),14-21.  R. Pizzi, O. Durin, G. Inama, 2008, Non-invasive assessment of risk for severe tachyarrhythmias by means of non-linear analysis techniques, in: Developments in Chaos and Complexity Research, Nova Science NY.  G. Inama, C. Pedrinazzi, O. Durin, M. Nanetti, G. Donato, R. Pizzi, 2007, Ventricular arrythmias in competitive atheltes: risk stratification with T-wave alternans, Heart International, 3(1),58-67,  H. Heidbüchel, J. Hoogsteen, R. Fagard R, 2003, High prevalence of right ventricular involvement in endurance athletes with ventricular arrhythmias. Role of an electrophysiologic study in risk stratification, Eur Heart J, 24(16), 1473-1480.  H. L. Kennedy, 2002, Ventricular ectopy in athletes. Do not worry ..... More good news, J. Am., Coll. Cardiol, 40(3), 453-456.  D. Corrado, C. Basso, 2003, Does sports activity enhance the risk of sudden death in adolescent and young adults?, J.Am. Coll. Cardiol. 42(11), 1959-1963.  P. Palatini, C. Miraglino, 1985, Prevalence and possible mechanisms of ventricular arrhythmias in athletes, Am Heart American Journal of Biomedical Engineering 2014, 4(3): 60-67 67 J. 110(3), 560-567. 5/1043.full.  A. Biffi, A. Pelliccia, 2002, Long-term clinical significance of  J. E. Mietus, C.K. Peng, I. Henry, R. L. Goldsmith, A. L. frequent and complex ventricular tachyarrhytmias in trained Goldberger, 2002, The pNNx files: re-examining a Widely athletes, J.Am.Coll.Cardiol, 40(3), 446-452. used heart rate variability measure, Heart, 88(4), 378-380.  A. Biffi, L. Verdile, 2004, Arrhythmias and relationship to  M. Costa, A. L. Goldberger, C. Peng, 2002, Multiscale physical training in athletes, From Arrhythmias in athletes Ital entropy analysis of physiologic time series, Phys Rev Lett, Heart J. Supp., 5(1), 143S - 144S. 89(62), 1-4.  F. Furlanello, R. Bettini, A. Bertoldi, G. Inama, 1989,  V. Batagelj, 1988, Generalized Ward and related clustering Arrhythmia patterns in athletes with arrhythmogenic right problems, in Classification and Related Methods of Data ventricular dysplasia, Eur Heart J. ,10(D), 6-19. Analysis, North-Holland, 67-74.  Organizing Committee for the Fitness for Sports Cardiology ANCE-ANMCO-FMSI-SIC-SICSPORT, 2009, Protocols for cardiological assessment of their suitability to competitive sport 2009. Rome, International Scientific Publishing House  T. Hastie, R. Tibshirani, J. Friedman, 2009, Hierarchical clustering in: The Elements of Statistical Learning: Data Mining, Inference, and Prediction, Second Edition, Ed. New York: Springer. 520–528.,  A. Pelliccia, R. Fagard, H. H. Bjørnstad, 2005, Study Group of Sports Cardiology of the Working Group of Cardiac Rehabilitation and Exercise Physiology; Working Group of Myocardial and Pericardial Diseases of the European Society of Cardiology, Recommendations for competitive sports participation in athletes with cardiovascular disease: a consensus document from the Study Group of Sports Cardiology of the Working Group of Cardiac Rehabilitation and Exercise Physiology and the Working Group of Myocardial and Pericardial Diseases of the European Society of Cardiology,” Eur Heart J. , 26(14), 1422-1445.  D. P. Zipes, A. J. Camm, M. Borggrefe, 2006, ACC / AHA / ESC 2006 guidelines for management of patients with ventricular arrhythmias and the prevention of sudden cardiac death, J Am Coll Cardiol, 48(5), 1064-1108.  F. Giada, G. Inama, M. Bertaglia, 2010, Management of patients with palpitations,” Consensus document of the AIAC (Italian Association of Arrhythmiology and Cardiostimulation), G Ital Cardiol, 11(4), 329-340.  The Task Force of the European Society of Cardiology the North American Society of Pacing Electrophysiology, 1996, Heart Rate Variability Standards of Measurement, Physiological Interpretation, and Clinical Use, Circulation, 93, 1043-1065, Available: http://circ.ahajournals.org/content/93/  A. D. Gordon, 1999, Classification, Second Edition, London Chapman and Hall / CRC 1999.  J. H. Ward, Jr. 1963, Hierarchical grouping to optimize an objective function, J. Am Stat. Ass, 58(301), 236–244.  R. Pizzi, M. de Curtis, C. Dickson, 2003, The Binding Problem: Evidence of Chaotic Attractors in Cortical Fast Oscillations Tested by an Artificial Neural Network, in: Advances in Soft Computing, Physica Springer-Verlag,  T. Kohonen, 1983, Self-Organisation and Association Memory: 3rd edition, Springer-Verlag NY.  T. Kohonen, 1993, Physiological Interpretation of the Self-Organizing Map Algorithm, J. Neural Net., 6(7), 895-905.  A. L. Goldberger, L. A. N. Amaral, L. Glass, J. M. Hausdorff, P. Ch. Ivanov, R. G. Mark, J. E. Mietus, G. B. Moody, C. K. Peng, H. E. Stanley, 2000, PhysioBank, PhysioToolkit, and PhysioNet: Components of a New Research Resource for Complex Physiologic Signals, Circulation, 101(23), e215-e220.  R Development Core Team, 2005, R: A language and environment for statistical computing, [Online] R Foundation for Statistical Computing, http://www.R-project.org.
... pages left unread,continue reading
Free reading is over, click to pay to read the rest ... pages
0 dollars，0 people have bought.
Reading is over. You can download the document and read it offline
0people have downloaded it