No abstract

Introduction: The aim of the Physionet/CinC Challenge 2017 is to automatically classify atrial fibrillation (AF) from a short single lead ECG recording. The Challenge provides 8,528 labeled ECG recordings; each recording was labeled as normal, AF, other, or noisy. In addition, the Challenge provides sample code which includes an R-peak detector and a simple classifier.Algorithm: We use an ensemble of features extracted from the ECG signals to create a four-class support vector machine (SVM) classifier. Included in the feature set are statistics obtained from the ECG signal, its spectrum, and the RR-intervals. In addition, we learn a 32-element sparse coding dictionary on the sorted RR-intervals of the ECG signals. Using the dictionary, we calculate a sparse coefficient vector for each training sample and put these through a soft-margin linear SVM. The soft-margin scores are used as additional features in the final classifier.Results: Our algorithm achieves cross-validated F1 scores of 0.874, 0.756, and 0.689 (for normal, AF, and other files, respectively), resulting in a final crossvalidated challenge score of 0.773. The score when tested on a subset of the unknown data is 0.78 (with F1 scores of 0.88, 0.80, 0.65). The official challenge score was 0.77.Conclusions: We developed an algorithm to classify ECG recordings as normal, AF, other, or noisy. Our results show that sparse coding is an effective way to define discriminating features from a list of sorted RR-intervals. In addition, these sparse codes complement more commonly used features in the classification task. Further work will attempt to increase the accuracy of the algorithm by exploring other features and classifiers while still using sparse coding as an unsupervised feature extractor.

Background: Despite the limited donor pool, current guidelines recommend against the use of hearts from donors that abuse alcohol. Study aim: To explore the effect of donor alcohol abuse (AA) on cardiac allograft arrhythmias, function, and outcomes in heart transplant (HTx) recipients. Methods: Overall, 370 HTx recipients transplanted between 2005 and 2016 were divided into two groups: (1) the AA donor group (AD, n=58), and (2) the non-AA donor group (NAD, n=312). The median follow-up was 4.0 (25%-75% IQR 1.8-5.8) years. Results: Recipients in the AD group had a slower heart rate (HR, 86 §13 vs. 93 §13, p=0.004) and an increased incidence of early atrial fibrillation [17 (30%) vs. 34 (11%), p=0.003]. Echocardiographic left ventricular mass was higher in AA donors (171.7 §66.7 vs. 151.6 §54.7, p=0.02). This difference became even more prominent 1 year following HTx (185 §43 vs. 166 §42, p=0.007). E/E' was higher in the AD group (9.5 §3.9 vs. 8.4 §2.9, p=0.04) and a larger number of AD recipients had a ventilatory equivalent for VCO 2 >34 (50% vs. 31%, p=0.04) on cardiopulmonary stress test. There was no significant difference in rejection, cardiac allograft vasculopathy (CAV), or survival between the groups. Conclusions: Our data suggest that donor AA is associated with an increased incidence of post-HTx atrial fibrillation and impaired cardiac allograft diastolic function. Therefore, although the incidence of rejection, CAV, and intermediate-term survival were not affected, hearts from donors that abuse alcohol should be used with caution.Background: Ambulatory home monitoring of heart failure (HF) patients (pts) to assess clinical status remotely and adjust therapies accordingly could potentially decrease the costs associated with hospitalizations and improve pts quality of life. Our group has been investigating the use of non-invasive ballistocardiogram (BCG) measurements (the measurement of body vibrations due to cardiac ejection of blood) using a modified weighing scale for monitoring hemodynamics and cardiac timing intervals from HF pts. BCG waveforms from HF pts in a decompensated state will show, on average, a higher variability in signal structure than corresponding waveforms from pts in a compensated state. Methods: For each of the 36 HF pts included in the study, we recorded simultaneously daily 30-second measurements of electrocardiogram (ECG) and BCG signals using a modified weighing scale (Fig. 1a). Recordings were processed for data quality and some of the waveforms were discarded if data quality was insufficient (e.g. missing R-peaks in ECG signals). We developed a metric from the BCG signal that quantifies the variability of the BCG heartbeat waveforms (Fig. 1b). Results: Of the 36 pts (83% men, 58 § 13 years, left ventricular ejection fraction 0.31 § 0.12, New York Heart Association functional class I/ II 36% and III 64%), 18 pts were recorded only when hospitalized for a HF