Diagnostic decision support systems for atrial fibrillation based on a novel electrocardiogram approach.

BACKGROUND: The electrocardiogram (ECG) is one of the most non-invasive techniques to give support to the atrial fibrillation (AF) diagnosis. Several authors use the temporal difference between two consecutive R waves, a method known as RR interval, to perform the AF diagnosis. However, RR interval-based analysis does not detect distortions on the other ECG waves. PURPOSE: Thus, the present work proposes a diagnostic decision support systems for AF based on higher order spectrum analysis of the voltage variation on the ECG.. METHODS: The proposed method was used aiming AF classifying. The classifier is composed by two screening stages: one based on the average and another on the average deviation of kurtosis of the ECG signals. Heartbeat obtained from the MIT-BIH atrial fibrillation and MIT-BIH normal were used. RESULTS: ECG signal featured by kurtosis outperforms second order statistics based metrics in up to 476 times, and up to 110 times above the RR interval. The screening methods obtained sensitivity equal to 100% and specificity is up to 84.04%. The two screening methods combined provided an AF classifier with an accuracy rate at diagnosis of 100%. The results presented take into account windows of up to five heartbeats and a 99.73% confidence interval. CONCLUSION: The results obtained by the proposed method can be used to support decision-making in clinical practices with a diagnostic accuracy rate of 90.04% to 100%.

The Performance of Short-Term Heart Rate Variability in the Detection of Congestive Heart Failure.

Congestive heart failure (CHF) is a cardiac disease associated with the decreasing capacity of the cardiac output. It has been shown that the CHF is the main cause of the cardiac death around the world. Some works proposed to discriminate CHF subjects from healthy subjects using either electrocardiogram (ECG) or heart rate variability (HRV) from long-term recordings. In this work, we propose an alternative framework to discriminate CHF from healthy subjects by using HRV short-term intervals based on 256 RR continuous samples. Our framework uses a matching pursuit algorithm based on Gabor functions. From the selected Gabor functions, we derived a set of features that are inputted into a hybrid framework which uses a genetic algorithm and k-nearest neighbour classifier to select a subset of features that has the best classification performance. The performance of the framework is analyzed using both Fantasia and CHF database from Physionet archives which are, respectively, composed of 40 healthy volunteers and 29 subjects. From a set of nonstandard 16 features, the proposed framework reaches an overall accuracy of 100% with five features. Our results suggest that the application of hybrid frameworks whose classifier algorithms are based on genetic algorithms has outperformed well-known classifier methods.

Statistical coding and decoding of heartbeat intervals.

The heart integrates neuroregulatory messages into specific bands of frequency, such that the overall amplitude spectrum of the cardiac output reflects the variations of the autonomic nervous system. This modulatory mechanism seems to be well adjusted to the unpredictability of the cardiac demand, maintaining a proper cardiac regulation. A longstanding theory holds that biological organisms facing an ever-changing environment are likely to evolve adaptive mechanisms to extract essential features in order to adjust their behavior. The key question, however, has been to understand how the neural circuitry self-organizes these feature detectors to select behaviorally relevant information. Previous studies in computational perception suggest that a neural population enhances information that is important for survival by minimizing the statistical redundancy of the stimuli. Herein we investigate whether the cardiac system makes use of a redundancy reduction strategy to regulate the cardiac rhythm. Based on a network of neural filters optimized to code heartbeat intervals, we learn a population code that maximizes the information across the neural ensemble. The emerging population code displays filter tuning proprieties whose characteristics explain diverse aspects of the autonomic cardiac regulation, such as the compromise between fast and slow cardiac responses. We show that the filters yield responses that are quantitatively similar to observed heart rate responses during direct sympathetic or parasympathetic nerve stimulation. Our findings suggest that the heart decodes autonomic stimuli according to information theory principles analogous to how perceptual cues are encoded by sensory systems.