Genetic Algorithms to Simplify Prognosis of Endocarditis

Abstract

This ongoing interdisciplinary research is based on the application of genetic algorithms to simplify the process of predicting the mortality of a critical illness called endocarditis. The goal is to determine the most relevant features (symptoms) of patients (samples) observed by doctors to predict the possible mortality once the patient is in treatment of bacterial endocarditis. This can help doctors to prognose the illness in early stages; by helping them to identify in advance possible solutions in order to aid the patient recover faster. The results obtained using a real data set, show that using only the features selected by employing a genetic algorithm from each patient's case can predict with a quite high accuracy the most probable evolution of the patient.

Clustering avatars behaviours from virtual worlds interactions

“…These methods are usually known as feature selection methods [19]. Curiel et al [11] applied Genetic Algorithms [16] to simplify prognosis of endocarditis using a codification where each individual of the population is a feature set. Blum and Langley [7] showed some examples of relevant features selection in datasets and applied them to several Machine Learning techniques.…”