Publication:

Implementación de una arquitectura basada en plataformas de computación de alto desempeño y tiempo real para el procesamiento de señales cardíacas

E-Health and M-Health represent a great opportunity to overcome the big limitations of coverage, equity, and quality of health services in developing countries. M-Health solutions that aim to the continuous monitoring and analysis of heart signals are particularly promising, considering that heart diseases are among the top 3 causes of death in the world. However, deploying this kind of solutions at a region-wide scale, guaranteeing -among other quality attributes- high availability and good performance is a big challenge. A signal processing platform working at this scale requires processing continuously and simultaneously signals from thousands of patients. Moreover, those platforms are deployed over complex, heterogeneous and dynamic topologies, where millions of devices exchange significant amounts of information over global communications infrastructures. In this paper, we present a highly scalable architecture that integrates Real-Time and High-Performance Computing platforms (used for BigData and GPU computing respectively) to minimize the latency of the processing time of a massive amount of signal streams. We validate our approach through a prototype implementation that integrates real ECG processing algorithms. According to the preliminary experiments and results, with the proposed architecture the platform should be capable of processing thousands of concurrent ECG signals, efficiently calculating the heart rate, heart rate variability, and several cardiopathies in real-time.