Especificação e Monitoramento de Requisitos de Qualidade de Contexto para Aplicações de Fenotipagem Digital

Abstract

Digital phenotyping applications use sensor data from personal digital devices (smartphones, smart bands) to quantify the moment-to-moment human phenotype at the individual in-situ level. Some aspects can degrade the Quality of Context (QoC) of applications, such as inaccurate sensor information, wireless communication technologies used in acquiring and distributing information, scalability problems, and intermittent connection due to user mobility. Ensuring the quality and distribution of the data used is an essential requirement in the domain of these applications. Therefore, this study conceived a process to incorporate QoC requirements in applications for digital phenotyping. The process consists of five steps, namely: (i) specifying the QoC requirements using a metamodel; (ii) target code generation through a transformation mechanism that receives the requirements specification as input; (iii) the incorporation in the application of the generated source codes to guarantee the QoC requirements; (iv) evaluation and monitoring of QoC parameters; (iv) visualization of the monitoring result through Dashboards, which aims to provide the developer with an environment in which it is possible to monitor the quality level of the application instances. A case study was developed to evaluate the applicability of the proposed process in a typical application of digital phenotyping. Additionally, experiments were carried out to analyze the application’s performance with and without the QoC mechanisms to evaluate the impact of the QoC during its execution. By applying the process proposed in the developed case study, we observed the ease of incorporating QoC requirements in applications for digital phenotyping. When carrying out the QoC experiments, it was possible to verify that the application instrumented with the QoC mechanisms guarantees the quality and delivery of the information that will be used for the inference of situations of interest.