A hybrid decision support system for the identification of asthmatic subjects in a cross-sectional study

This paper discusses the implementation of a decision support system for the prediction of asthma in a group of children with related medical factors. The system makes use of the survey data that is gathered as part of ISAAC Phase One Study, obtained through questionnaires completed by adolescents at school and at home by the parents of the children. The model is tested on cross-sectional study data that involves two age groups. The decision support system is basically hybrid in nature as it involves the fusion of unsupervised and supervised learning techniques. As part of preprocessing, feature clustering is performed to identify the features that have a high degree of correlation with the asthma feature in the cluster. Further,this information is used to identify subject clusters by applying modified Fuzzy C Means Clustering. Clustering results are evaluated using Silhouette values. A decision tree is further constructed by using this information which in turn is used to predict the presence or absence of asthma by deploying regression.The performance of the overall model is estimated by analyzing sensitivity and specificity for the obtained prediction results which is quite satisfactory.