A Probabilistic Model for COPD Diagnosis and Phenotyping Using Bayesian Networks

Introduction: This research was meant to provide a model for COPD diagnosis and to classify the cases into phenotypes; General COPD, Chronic bronchitis, Emphysema, and the Asthmatic COPD using a Bayesian Network (BN). Methods: The model was constructed through developing the Bayesian Network structure and instantiating the parameters for each of the variables. In order to validate the achieved results, the same data set was applied to a neural network application using the Levenberge- Marquardt algorithm. Furthermore, a card Diag, a C++ application that enables graphical classification of COPD into phenotypes and depicts the relationships of COPD phenotypes was developed. Results: The results showed that a Bayesian Network can be successfully applied to develop a probabilistic model for diagnosis and classification of COPD cases into the corresponding phenotypes. Conclusions: A model that classifies COPD cases into phenotypes of general COPD, Chronic bronchitis, Emphysema, and Asthmatic COPD was successfully developed. Moreover, the achieved results also helped to represent graphical representations of COPD phenotypes and explained how the phenotypes relate to each other. It was also observed that COPD is mostly associated with people aged 40 years or older. Overall, smoking is the major cause of COPD.