ChronicPrediction: um modelo para prognóstico ubíquo de fatores de risco de doenças crônicas não transmissíveis

Abstract

The ubiquitous computing in the form os ubiquitous systems and used in the support and care of Chronic Diseases prioritize the patient monitoring and the generation of differents alert types, however, the support decision making by the existing ubiquitous systems is still little used on specific systems for the management and control of Chronic Non-Communicable Diseases. As the care of chronic disease should be done continuosly, becomes important for the patient has a prior knowledge about the progress of your treatment and if the actions taken by him in his daily life are helping you with treatment or not. As a predictive mechanism one of the main techniques used nowadays are the Bayesian Networks. Thus, this thesis proposes an ubiquitous computing prognostic model of risk factors of Chronic Noncommunicable Diseases, called ChronicPrediction. The ChronicPrediction model uses Bayesian Networks created from mapping of existing causal relationships between each of the risk factors of NCDs which you wish to observe. These reationships are defined from expert opinion or automatically generated by historical data and based on data provided by patients themselves about their dayli eating habits, exercise routine and the measuring of their rates. Are also discussed characteristics belonging to related work, addition to describing the model in detail and present the aspects considered in developing and evaluating through a prototype. The evaluation process is presented in the form of experiments described through scenarios, which have to evaluate hypotheses realted to each. The starting point for the formulation of each of the hypotheses is the fact that we have an idea of a cause and effect related to it. Each scenario aims to describe common situations that may occur during the daily lives of patients (causes and effects) with some kind of Chronic Non-Communicable Disease. Furthermore, the diversity between the scenarios is important to improve the coverage of the model evaluation. Making the evaluationsit was concluded that the ChronicPrediction model expands the functionality of UDuctor model and the ChronicDuctor personal assistant, offering support to the monitoring of multiple NCDs simultaneously, providing feedbacks and recommendations to the patients in order to help them to monitor their treatment continuously, to modify them in order to promote their well-being and improving their quality of life.