Abstract
Auscultation is a widely used technique in clinical activity to diagnose heart diseases. However, heart sounds are difficult to interpret because a) of events with very short temporal onset between them (tens of milliseconds) and b) dominant frequencies that are out of the human audible spectrum. In this paper, we propose a model to segment heart sounds using a semi-hidden Markov model instead of a hidden Markov model. Our model in difference from the state-of-the-art hidden Markov models takes in account the temporal constraints that exist in heart cycles. We experimentally confirm that semi-hidden Markov models are able to recreate the "true" continuous state sequence more accurately than hidden Markov models. We achieved a mean error rate per sample of 0.23.

Abstract
Heart sounds are difficult to interpret because a) they are composed by several different sounds, all contained in very tight time windows; b) they vary from physiognomy even if the show similar characteristics; c) human ears are not naturally trained to recognize heart sounds. Computer assisted decision systems may help but they require robust signal processing algorithms. In this paper, we use a real life dataset in order to compare the performance of a hidden Markov model and several hidden semi Markov models that used the Poisson, Gaussian, Gamma distributions, as well as a non-parametric probability mass function to model the sojourn time. Using a subject dependent approach, a model that uses the Poisson distribution as an approximation for the sojourn time is shown to outperform all other models. This model was able to recreate the "true" state sequence with a positive predictability per state of 96%. Finally, we used a conditional distribution in order to compute the confidence of our classifications. By using the proposed confidence metric, we were able to identify wrong classifications and boost our system (in average) from an approximate to 83% up to approximate to 90% of positive predictability per sample.

Abstract
This paper, presents ongoing work that extends MetaProbLog with Most Probable Explanation (MPE) inference method. The MPE inference method is widely used in Hidden Markov Models in order to derive the most likely states of a model. Recently, we started developing an application that uses MetaProbLog to models phonocardiograms. We target to use this application in order to diagnose heart diseases by using phonocardiogram classification. Motivated by the importance of phonocardiogram classification, we started the implementation of the MPE inference method and an improvement of representation for annotated disjunctions.

Abstract
Recently, soft attention mechanisms have been successfully used in a wide variety of applications such as the generation of image captions, text translation, etc. This mechanism attempts to mimic the visual cortex of a human brain by not analyzing all the objects in a scene equally, but by looking for clues (or salient features) which might give a more compact representation of the environment. In doing so, the human brain can process information more quickly and without overloading. Having learned this lesson, in this paper, we try to make a bridge from the visual to the audio scene classification problem, namely the classification of heart sound signals. To do so, a novel approach merging soft attention mechanisms and recurrent neural nets is proposed. Using the proposed methodology, the algorithm can successfully learn automatically significant audio segments when detecting and classifying abnormal heart sound signals, both improving these classification results and somehow creating a simple justification for them.

Abstract
The analysis of heart sounds is a challenging task, due to the quick temporal onset between successive events and the fact that an important fraction of the information carried by phonocardiogram (PCG) signals lies in the inaudible part of the human spectrum. For these reasons, computer-aided analysis of the PCG can dramatically improve the quantity of information recovered from such signals. In this paper, a hidden semi-Markov model (HSMM) is used to automatically segment PCG signals. In the proposed models, the emission probability distributions are approximated via Gaussian mixture model (GMM) priors. The choice of GMM emission probability distributions allow to apply re-estimation routines to automatically adjust the HSMM emission probability distributions to each subject. Building on the proposed method for fine tuning emission distributions, a novel subject-driven unsupervised heart sound segmentation algorithm is proposed and validated over the publicly available PhysioNet dataset. Perhaps surprisingly, the proposed unsupervised method achieved results in line with state-of-the-art supervised approaches, when applied to long heart sounds.

Abstract
The first step in any non linear time series analysis, is to characterize signals in terms of periodicity, stationarity, linearity and predictability. In this work we aim to find if PCG (phonocardiogram) and ECG (electrocardiogram) time series are generated by a deterministic system and not from a random stochastic process. If PCG and ECG are non-linear deterministic systems and they are not very contaminated with noise, data should be confined to a finite dimensional manifold, which means there are structures hidden under the signal that could be used to increase our knowledge in forecasting future values of the time series. A non-linear process can give rise to very complex dynamic behaviours, even though the underlying process is purely deterministic and probably low-dimensional. To test this hypothesis, we have generated 99 surrogates and then we compared the fitting capability of AR (auto-regressive) models on the original and surrogate data. The results show with a 99\% of confidence level that PCG and ECG were generated by a deterministic process. We compared the fitting capability of an ECG and PCG to AR linear models, using a multi-channel approach. We make an assumption that if a signal is more linearly predictable than another one, it may adjust better to these AR linear models. The results showed that ECG is more linearly predictable (for both channels) than PCG, although a filtering step is needed for the first channel. Finally we show that the false nearest neighbour method is insufficient to identify the correct dimension of the attractor in the reconstructed state space for both PCG and ECG signals.