Medical Monkeys: A Crowdsourcing Approach to Medical Big Data

Abstract

Big data play a central role in eHealth and have been crucial for designing and implementing clinical decisions support systems. Those applications can avail on data analysis and response capabilities, often empowered by Machine Learning algorithms, which can help clinician in diagnostic as well as therapeutic decisions. On the other hand, in the context of eSociety, eCommunities can be essential actors for managing and structuring medical data. In fact, they can support in gathering, providing and labeling data. This last task is highly relevant for medical Big Data, as it is a key point for supervised Machine Learning algorithms, which need an extensive data annotation process. This improves prediction and analysis capabilities of the algorithms on large datasets. Our approach on the medical Big Data labeling problem is the design and prototyping of a crowdsourcing collaborative Web Application, used for the annotation of medical images, that we named Medical Monkeys. Under the principles of mutual advantage and collaboration researchers, online gamers, medical students and patients will be involved, within this platform, in a virtual and mutually beneficial cooperation for improving Machine Learning algorithms. Using our application on large scale data analysis, algorithms for image segmentation will become useful for clinical decisions support systems. Our application is the result of a collaboration of several universities and research institutes and has, as principal aim, the integration, in form of gaming tasks, of eCommunities for the implementation of a more accurate analysis and diagnostic on MRI or CT images.