“Making sense out of multi-channel physiological data for pervasive health applications”. Danilo Mandic, Sithan Kanna, Andrzej Cichocki.

The emergence of cheap but limited resolution physiological sensors which will be embedded in next-generation wearable devices presents new opportunities for the use of statistical signal processing tools in real-time and continuous health monitoring applications. Indeed, signal processing in this context (data conditioning, detection, estimation, separation) promises to provide deeper insights into the information-bearing components within these multi-channel measurements. Unlike in traditional signal processing applications, the signals obtained from body sensor networks often have very low signal-to-noise ratios and may contain large artefacts caused by motion or muscle contractions, such as in notoriously noisy Electroencephalogram (EEG) monitoring. Moreover, embedding the data processing capabilities within ultra-wearable miniature devices can also prove to be a challenge wherein the trade-offs between the accuracy of the algorithm, computational complexity and power requirements have to be considered.

These issues pose significant challenges that need to be addressed by the statistical signal processing community for pervasive and real-time health monitoring to become a reality. To this end, the special session aims to bring together the latest advances in statistical signal processing research applied to multichannel physiological data, focusing on real- world applications for next-generation personalised healthcare. Special attention will be paid to the algorithms that are intended for use in the diagnosis of disease or other conditions, or in the cure, mitigation, treatment, or prevention of disease.