ABSTRACT
The analysis of cardiac vibration signals has been shown as an interesting tool for the follow-up of chronic pathologies involving the cardiovascular system, such as heart failure (HF). However, methods to obtain high-quality, real-world and longitudinal data, that do not require the involvement of the patient to correctly and regularly acquire these signals, remain to be developed. Implantable systems may be a solution to this observability challenge. In this paper, we evaluate the feasibility of acquiring useful electrocardiographic (ECG) and accelerometry (ACC) data from an innovative implant located in the gastric fundus. In a first phase, we compare data acquired from the gastric fundus with gold standard data acquired from surface sensors on 2 pigs. A second phase investigates the feasibility of deriving useful hemodynamic markers from these gastric signals using data from 4 healthy pigs and 3 pigs with induced HF with longitudinal recordings. The following data processing chain was applied to the recordings: (1) ECG and ACC data denoising, (2) noise-robust real-time QRS detection from ECG signals and cardiac cycle segmentation, (3) Correlation analysis of the cardiac cycles and computation of coherent mean from aligned ECG and ACC, (4) cardiac vibration components segmentation (S1 and S2) from the coherent mean ACC data, and (5) estimation of signal context and a signal-to-noise ratio (SNR) on both signals. Results show a high correlation between the markers acquired from the gastric and thoracic sites, as well as pre-clinical evidence on the feasibility of chronic cardiovascular monitoring from an implantable cardiac device located at the gastric fundus, the main challenge remains on the optimization of the signal-to-noise ratio, in particular for the handling of some sources of noise that are specific to the gastric acquisition site.
ABSTRACT
Advancement on computer and sensing technologies has generated exponential growth in the data available for the development of systems that support decision-making in fields such as health, entertainment, manufacturing, among others. This fact has made that the fusion of data from multiple and heterogeneous sources became one of the most promising research fields in machine learning. However, in real-world applications, to reduce the number of sources while maintaining optimal system performance is an important task due to the availability of data and implementation costs related to processing, implementation, and development times. In this work, a novel method for the objective selection of relevant information sources in a multimodality system is proposed. This approach takes advantage of the ability of multiple kernel learning (MKL) and the support vector machines (SVM) classifier to perform an optimal fusion of data by assigning weights according to their discriminative value in the classification task; when a kernel is designed for representing each data source, these weights can be used as a measure of their relevance. Moreover, three algorithms for tuning the Gaussian kernel bandwidth in the classifier prediction stage are introduced to reduce the computational cost of searching for an optimal solution; these algorithms are an adaptation of a common technique in unsupervised learning named local scaling. Two real application tasks were used to evaluate the proposed method: the selection of electrodes for a classification task in Brain-Computer Interface (BCI) systems and the selection of relevant Magnetic Resonance Imaging (MRI) sequences for detection of breast cancer. The obtained results show that the proposed method allows the selection of a small number of information sources.
