Private Data Analytics on Biomedical Sensing Data Via Distributed Computation

Private Data Analytics on Biomedical Sensing
Data Via Distributed Computation

Abstract— Firstly advances in biomedical sensors, mobile communication technologies have protected fast growth of mobile health applications. Users generate a high size biomedical data during health monitoring. Therefore it can be used by mHealth server for training predictive models. In addition disease finding and treatment. However the data raise serious privacy dealings.Because they report exact information. Probably health status, lifestyles. While proposes and experimentally studies an idea that keeps training samples private. Furthermore enabling accurately constructs predictive models. As a result, specifically consider logistic types. Especially widely used for imagine dichotomous ends in healthcare, turn logistic regression problem into small sub problems. Hence done over two types of distributed sensing data. That are horizontally and vertically partitioned. Finally the sub problems are solved using individual private input. Thus mHealth users keep their data locally. Also can only upload intermediate results to the mHealth server for model training < Final Year Projects 2016 > Practical results based on real datasets show that our plan is highly useful and accessible to a large number of users.