Asthma Disease Prediction Using Regression Tree Method

Asthma is a common, chronic inflammatory disorder of the airways that affects an estimated 339 million people worldwide. Diagnostic approaches for asthma usually fail in clinical practice, partly owing to the multifactorial spectrum of the disease. This study reveals a new diagnostic algorithm where regression trees along with entropy-based subset selection(E-SS) are combined for more reliable and accurate asthma diagnosis. E-SS helps to filter out the most important features from high-dimensional datasets and avoids possible overfitting of the rate up to 91.304%, which is higher than other algorithms like Bayesian Network of 83.3%. The strength of this model is that it can capture complex (non-linear) interactions efficiently between the variables and therefore would be efficient, in particular for asthma prediction. Moreover, it is a more patient-centered methodology where risk factors of each individual are targeted. The model could aid in the diagnosis and treatment of other chronic diseases outside asthma, further alleviating global health care systems.