Uma abordagem baseada em linguagens específicas de domínio e transformação de modelos para a geração semiautomática de aplicações mHealth para escores clínicos

Abstract

In medicine, clinical scores are tools designed to predict clinical outcomes, perform risk stratification, assist in clinical decision making, assess the severity of a disease, or aid diagnosis. These tools generally work in the context of chronic diseases, and are structured by a section of variables, a section of rules for calculating score scores, and an evaluation section. Recently, initiatives have emerged within Mobile Health to automatically resolve clinical score issues. These approaches are limited to developing a computational system for each score specificity. In this scenario, productivity is affected because every time we need to deploy an application to a new score, all steps of a traditional software development process will be repeated. In addition, the intrinsic complexity of the technological environment surrounding such applications requires specialized knowledge of various areas of medicine and computer science. This dissertation presents a new approach to software development focused on the automation of clinical scores in ubiquitous computing environments called MDD4ClinicalScores. MDD4ClinicalScores provides semi-automatic generation of software components through the use of metamodeling and model transformation. These components are submitted to the execution environment implemented by the approach for the creation of a mobile patient monitoring application based on the previously specified clinical score. The proposed approach was evaluated under two aspects: the expressiveness of the DSML4ClinicalScore language through the metamodeling of eight concrete cases, and the viability of the MDD4ClinicalScores through the development of two case studies.