Syndromes diagnostic model for coronary artery disease (CAD): An improved naïve Bayesian classification model based on attribute relevancy

Clinical decision support is a probabilistic and quantitative method learning used for problem modeling in situations of ambiguity. Nowadays, mathematical methods such as probability learning and statistical analysis is increasingly used in the medical field to assure the accuracy of the diagnoses and optimize the therapy. This study describes the design and implementation of an improved weighted Naïve Bayes classifier based on the idea that there is a certain level of dependency between symptoms. For the coronary artery disease, clinical data and patients' medical records were used to train the improved Naïve Bayes classifier to compute the possible disease manifestations, the priori probability and the posterior probability and to carry out the disease classification. Accordingly, we were able to improve the precision of the improved weighted Naïve Bayes classifier to 79% (increase of 12% compared to Naïve Bayes). The suitability of the improved Naïve Bayes classifiers for carrying out diagnosis allows them to be utilized to distinguish syndromes and signs based on unobserved evidence.