A WEB ARCHITECTURE OF A FRAMEWORK TO EVALUATE SMARTPHONE-BASED FALL DETECTION SYSTEMS.

Smartphones have become an interesting alternative to deploy wearable and cost-efficient Fall Detection Systems (FDSs). During last five years, many research works have proposed a variety of algorithms for Smartphone-based FDSs. These systems are typically evaluated by testing the detection decision with a set of predefined movements (emulated falls and Activities of Daily Living or ADLs). However, up to date, there is an almost complete lack of a 'benchmarking' procedure to compare the actual performance of these algorithms in a rigorous way. This paper proposes an architecture that enables investigators to create and share with the research community their own datasets with falls and ADLs. The architecture, which consists of a simple downloadable smartphone application and a database server that can be accessed from a conventional web browser, is aimed at configuring a global framework to assess the effectiveness of FDSs based on commonly accepted evaluation patterns. [ABSTRACT FROM AUTHOR]