Frida Sandberg

Frequency Tracking of Atrial Fibrillation Using Hidden Markov Models

A hidden Markov model (HMM) is employed to improve noise robustness when tracking the dominant frequency of atrial fibrillation (AF) in the electrocardiogram (ECG). Following QRST cancellation, a sequence of observed frequency states is obtained from the residual ECG, using the short-time Fourier transform. Based on the observed state sequence, the Viterbi algorithm retrieves the optimal state sequence by exploiting the state transition matrix, incorporating knowledge on AF characteristics, and the observation matrix, incorporating knowledge of the frequency estimation method and signal-to-noise ratio (SNR). The tracking method is evaluated with simulated AF signals to which noise, obtained from ECG recordings, has been added at different SNRs. The results show that the use of HMM improves performance considerably by reducing the rms error associated with frequency tracking: at 4-dB SNR, the rms error drops from 0.2 to 0.04 Hz.

An Atrioventricular Node Model for Analysis of the Ventricular Response During Atrial Fibrillation

This paper introduces a model of the atrioventricular node function during atrial fibrillation (AF), and describes the related ECG-based estimation method. The proposed model is defined by parameters that characterize the arrival rate of atrial impulses, the probability of an impulse choosing either one of the two atrioventricular nodal pathways, the refractory periods of these pathways, and the prolongation of the refractory periods. These parameters are estimated from the RR intervals using maximum likelihood estimation, except for the shorter refractory period which is estimated from the RR interval Poincaré plot, and the mean arrival rate of atrial impulses by the AF frequency. Simulations indicated that 200-300 RR intervals are generally needed for the estimates to be accurate. The model was evaluated on 30-min ECG segments from 36 AF patients. The results showed that 88% of the segments can be accurately modeled when the estimated probability density function (PDF) and an empirical PDF were at least 80% in agreement. The model parameters were estimated during head-up tilt test to assess differences caused by sympathetic stimulation. Both refractory periods decreased as a result of stimulation, and the likelihood of an impulse choosing the pathway with the shorter refractory period increased.

Classification of Paroxysmal and Persistent Atrial Fibrillation in Ambulatory ECG Recordings

The problem of classifying short atrial fibrillatory segments in ambulatory ECG recordings as being either paroxysmal or persistent is addressed by investigating a robust approach to signal characterization. The method comprises preprocessing estimation of the dominant atrial frequency for the purpose of controlling the subbands of a filter bank, computation of the relative subband (harmonics) energy, and the subband sample entropy. Using minimum-error-rate classification of different feature vectors, a data set consisting of 24-h ambulatory recordings from 50 subjects with either paroxysmal (26) or persistent (24) atrial fibrillation (AF) was analyzed on a 10-s segment basis; a total of 212,196 segments were classified. The best performance in terms of area under the receiver operating characteristic curve was obtained for a feature vector defined by the subband sample entropy of the dominant atrial frequency and the relative harmonics energy, resulting in a value of 0.923, whereas that of the dominant atrial frequency was equal to 0.826. It is concluded that paroxysmal and persistent AFs can be discriminated from short segments with good accuracy at any time of an ambulatory recording.

Atrioventricular nodal function during atrial fibrillation: Model building and robust estimation

Statistical modeling of atrioventricular (AV) nodal function during atrial fibrillation (AF) is revisited for the purpose of defining model properties and improving parameter estimation. The characterization of AV nodal pathways is made more detailed and the number of pathways is now determined by the Bayesian information criterion, rather than just producing a probability as was previously done. Robust estimation of the shorter refractory period (i.e., of the slow pathway) is accomplished by a Hough-based technique which is applied to a Poincare plot of RR intervals. The performance is evaluated on simulated data as well as on ECG data acquired from AF patients during rest and head-up tilt test. The simulation results suggest that the refractory period of the slow pathway can be accurately estimated even in the presence of many artifacts. They also show that the number of pathways can be accurately determined. The results from ECG data show that the refined AV node model provides significantly better fit than did the original model, increasing from 85 +/- 5% to 88 +/- 4% during rest, and from 86 +/- 5% to 87 +/- 3% during tilt. When assessing the effect of sympathetic stimulation, the AF frequency increased significantly during tilt (6.25 +/- 0.58 Hz vs. 6.32 +/- 0.61 Hz, p <0.05, rest vs. tilt) and the prolongation of the refractory periods of both pathways decreased significantly (slow pathway: 0.23 +/- 0.20 s vs. 0.11 +/- 0.10 s, p <0.001, rest vs. tilt; fast pathway: 0.24 +/- 0.31 s vs. 0.16 +/- 0.19s, p <0.05, rest vs. tilt). The results show that AV node characteristics can be assessed noninvasively for the purpose of quantifying changes induced by autonomic stimulation

Noninvasive Techniques for Prevention of Intradialytic Hypotension

Episodes of hypotension during hemodialysis treatment constitutes an important clinical problem which has received considerable attention in recent years. Despite the fact that numerous approaches to reducing the frequency of intradialytic hypotension (IDH) have been proposed and evaluated, the problem has not yet found a definitive solution-an observation which, in particular, applies to episodes of acute, symptomatic hypotension. This overview covers recent advances in methodology for predicting and preventing IDH. Following a brief overview of well-established hypotension-related variables, including blood pressure, blood temperature, relative blood volume, and bioimpedance, special attention is given to electrocardiographic and photoplethysmographic (PPG) variables and their significance for IDH prediction. It is concluded that cardiovascular variables which reflect heart rate variability, heart rate turbulence, and baroreflex sensitivity are important to explore in feedback control hemodialysis systems so as to improve their performance. The analysis of hemodialysis-related changes in PPG pulse wave properties hold considerable promise for improving prediction.

Circadian variation in dominant atrial fibrillation frequency in persistent atrial fibrillation

Circadian variation in atrial fibrillation (AF) frequency is explored in this paper by employing recent advances in signal processing. Once the AF frequency has been estimated and tracked by a hidden Markov model approach, the resulting trend is analyzed for the purpose of detecting and characterizing the presence of circadian variation. With cosinor analysis, the results show that the short-term variations in the AF frequency exceed the variation that may be attributed to circadian. Using the autocorrelation method, circadian variation was found in 13 of 18 ambulatory ECG recordings (Holter) acquired from patients with long-standing persistent AF. Using the ensemble correlation method, the highest AF frequency usually occurred during the afternoon, whereas the lowest usually occurred during late night. It is concluded that circadian variation is present in most patients with long-standing persistent AF though the short-term variation in the AF frequency is considerable and should be taken into account.

Application of frequency and sample entropy to discriminate long-term recordings of paroxysmal and persistent atrial fibrillation

Atrial fibrillation (AF) is the most common arrhythmia in clinical practice. At an early stage of the disease, AF may terminate spontaneously and is then referred to as paroxysmal AF. On the other hand, when external intervention is required for the arrhythmia to terminate, it is referred to as persistent AF. In this work, a method to discriminate between paroxysmal and persistent AF in the long-term ECGs is presented. The dominant frequency as well as the organization of the atrial activity are employed to characterize AF. The dominant atrial frequency (DAF) is estimated using hidden Markov model based frequency tracking, and organization is estimated by the sample entropy of the main atrial wave (MAW) and the first two harmonics, respectively. Long-term variations in DAF and organization from 50 ECG recordings were evaluated, showing that episodes of paroxysmal AF were consistently associated with lower DAF and organization of the MAW and the harmonics, than was persistent AF. Discrimination of paroxysmal and persistent AF resulted in classification rates of 84.1±26.1%, thus suggesting that it possible to discriminate between paroxysmal and persistent AF in patients without previously known AF history.

Prediction of hypotension in hemodialysis patients

Intradialytic hypotension (IDH) is the most common adverse complication during hemodialysis. Its early prediction and prevention will dramatically improve the quality of life for patients with an end stage renal disease. In a recent study, changes in the normalized envelope of the test statistic of the photoplethysmograpic (PPG) signal were found to predict acute symptomatic IDH. In the present study, the PPG-based predictor is generalized to include a patient-dependent threshold which incorporates on-line information on heart rate variability and heart rate turbulence. From datasets with patients prone and resistant to IDH, the results show that symptomatic IDH could be correctly predicted in 9 out of 14 cases, while 5 out of 24 were falsely predicted. In a subset of the data containing only patients prone to IDH, acute symptomatic IDH could be correctly predicted in 5 out of 5 cases, with one false prediction out of 14. When testing the robustness of the predictor, no significant changes were observed in the test statistic when controlled changes occurred in dialysis fluid temperature, ultrafiltration rate and body position.

Extracting a Cardiac Signal From the Extracorporeal Pressure Sensors of a Hemodialysis Machine

Although patients undergoing hemodialysis treatment often suffer from cardiovascular disease, monitoring of cardiac rhythm is not performed on a routine basis. Without requiring any extra sensor, this study proposes a method for extracting a cardiac signal from the built-in extracorporeal venous pressure sensor of the hemodialysis machine. The extraction is challenged by the fact that the cardiac component is much weaker than the pressure component caused by the peristaltic blood pump. To further complicate the extraction problem, the cardiac component is difficult to separate when the pump and heart rates coincide. The proposed method estimates a cardiac signal by subtracting an iteratively refined blood pump model signal from the signal measured at the extracorporeal venous pressure sensor. The method was developed based on simulated pressure signals, and evaluated on clinical pressure signals acquired during hemodialysis treatment. The heart rate estimated from the clinical pressure signal was compared to that derived from a photoplethysmographic reference signal, resulting in a difference of 0.07 ± 0.84 beats/min. The accuracy of the heartbeat occurrence times was studied for different strengths of the cardiac component, using both clinical and simulated signals. The results suggest that the accuracy is sufficient for analysis of heart rate and certain arrhythmias.

Noninvasive estimation of organization in atrial fibrillation as a predictor of sinus rhythm maintenance.

Previous studies indicate that the predictive value of atrial fibrillatory rate in patients undergoing cardioversion of atrial fibrillation (AF) of long duration is limited. The present study investigates signal entropy in this setting. Standard 12-lead electrocardiograms (ECGs) were recorded from 66 consecutive patients with AF undergoing cardioversion and sample entropy estimated. Patients were followed for 4 weeks. At follow-up, 59% of the patients had relapsed to AF. The AF signal entropy of these patients before cardioversion was 0.099 ± 0.015, whereas it was 0.093 ± 0.012 among the 41% maintaining sinus rhythm (P = .02). As hypothesized, signal entropy was lower in patients who maintained sinus rhythm 4 weeks after cardioversion than in those who did not. However, the overlap was large, making its clinical value limited.