Feature Selection Using Intelligent Agents for Time Improvement in Medical Diagnosis Systems

Feature selection is an important task in medical applications, given that the dimensionality and numerosity of such datasets is very high. In these cases, time parameter becomes also important, along with classification accuracy, in estimating the performance of a leaning model. In this work we propose intelligent agent teams that are capable to automatically discover the best time to build models while keeping the general accuracy at highest levels. For computing attributes’ relevance for the classification process, several techniques were used: Information Gain, Gain Ratio, Cor-relation and Relief Attribute Evaluator. The two proposed agent teams discovered that an optimum subset composed by 20 attributes (out of 133 attributes of the initial dataset) leads to accuracy rates equally to the ones registered on the entire dataset, meaning 98%, using Naive Bayes learning model, while improving the time taken to build model from 0.1 seconds to 0.03 seconds. For the proposed dataset, Naïve Bayes outperformed other classification techniques, such as: J48, Random Forest, and Dl4MlpClassifier.The proposed agents also integrated the best discovered model into a chatbot that performs medical diagnosis based on the symptoms collected from users.