On medical informatics for pervasive and ubiquitous computing in eHealth

As the world moves towards the reality of “intelligent infrastructures,” many avenues open up for research on sensor - based intelligent and ubiquitous systems. Healthcare is one such application area, where sensors and mobile platforms are becoming more useful and hence the idea of analyzing the data feeds from sensors to extract useful meanings is gaining in popularity. Various data-mining techniques are used in this regard. Apart from these, stream processing and continuous event processing are also becoming popular. This paper is a broad survey article where we look into the emerging trends in Ubiquitous Healthcare Information Systems, especially, various approaches taken in order to successfully use data analytics techniques on the data streams coming from the sensors and mobile platforms to cluster patients into similar groups, or analytics processing on streaming data to detect abnormal medical conditions as early as possible. The paper also refers to a recent research on non-parametric classification of data, which has the potential to discover interesting patterns within physiological data, which may otherwise remain undetected and advocates it's case in the health domain. Considering the size of the population and hence the volume of data, there are several architectural challenges such as scalability and availability of the platforms and handling of “big-data.” We try to summarize how these problems have been addressed and whether the solutions are adequate or not.