A predictive framework for the assessment of asthma control level

Abstract

Asthma is a chronic respiratory disease that is reversible in nature and hence identification of the level of control on the disease can be an important intervention to reduce the morbidity and mortality of the disease. We propose a predictive framework that efficiently predicts the asthma control levels in patients by identifying cells and cytokines in bronchoalveolar lavage (BAL) that contribute significantly to the differences in the controls. We apply various regularized regression techniques to infer the best performing technique on the dataset under consideration. Further, a two class classification problem to distinguish controlled and uncontrolled asthma subjects was handled by deploying binary classifiers and the best performing classifier was adopted. The framework involved the application of feature scoring techniques to identify the risk factors. The work is validated on the data containing subjects including healthy, controlled and uncontrolled subjects, acquired from the Department of Asthma, Allergy and Lung Biology, King’s College London School of Medicine, U.K. which was available on the Dryad repository.