Self Management of Asthma Disease Using Machine Learning Techniques

Abstract

The symptoms of asthma can alter. There is no existing cure for asthma. Asthma affects more than 25 million Americans today. This number includes more than a million children. Patients with asthma risk dying if they don't obtain therapy. A long-term symptom of asthma is a sudden worsening of symptoms that can be unpredictable and even fatal. In this work, we use supervised machine learning and the Random Forest method to achieve better outcomes in the self-management of asthma. Machine learning typically uses this random forest approach for classification and regression problems. In our research, we found that both probabilistic and discriminant classifiers were capable of producing outstanding accuracy (AUC>0.87) for early warning.