Rough set based intelligent approach for identification of H1N1 suspect using social media

摘要

Social media data offer unique challenges and opportunities for monitoring and surveillance of public health. The identification of epidemic suspect depends on doctor’s experience, symptoms and laboratory tests. Delay in identifying the beginning of infectious epidemic results in a big damage to a society. To handle the cases of epidemic effectively, a low-cost, accurate and timely diagnosis system is needed. An intelligent technique based on Rough set theory for identifying suspect of H1N1 using social media, has been presented in this paper. Classification of symptoms from the dataset has been performed using machine learning techniques. From the large number of symptom attributes mined from the dataset, H1N1 related symptom attributes, have been extracted. These extracted attributes contribute maximum to the decision-making process. Rough set theory has been used to evaluate significant attributes (symptoms) from symptom attribute set by generating reducts using indiscernibility relation. Identification of suspects is performed using significant conditional attributes and dependency rules generated from reducts. The utilization of presented social media based medical decision support system turn out to be an effective approach to assist government and health agencies in decision-making.