No Risk, More Fun! Automating Breach of Confidentiality Risk Assessment for Android Mobile Health Applications

Abstract

With the rapidly rising number of mobile health (mHealth) applications (apps), it is unfeasible to manually review mHealth apps for information privacy risks. One salient information privacy risk of mHealth apps are confidentiality breaches. We explore whether and how static code analysis is a feasible technology for app review automation. Evaluation of our research prototype shows that, on average, our prototype detected one breach of confidentiality risk more than human reviewers. Contributions are the demonstration that static code analysis is a feasible technology for detection of confidentiality breaches in mHealth apps, the derivation of eight generic design patterns for confidentiality breach risk assessments, and the identification of architectural challenges that need to be resolved for wide-spread dissemination of breach of confidentiality risk assessment tools. In terms of effectiveness, humans still outperform computers. However, we build a foundation for leveraging computation power to scale up breach of confidentiality risk assessments.