The prediction of postoperative morbidity in coronary artery bypass grafting using Naïve Bayes Classification and Bayes Factor

In the healthcare system, Coronary Artery Bypass Grafting (CABG) surgery is performed to treat coronary artery disease. Complications of CABG postoperative disease, which is called postoperative morbidity, affect the length of stay of the patient in the hospital. Therefore, predict the risk of patients experiencing postoperative morbidity is required to optimize the use of resources in the hospital. This study proposed about predicting the postoperative morbidity from patients who underwent CABG based on risk factors by using Bayes Factor and Naive Bayes Classification. The Bayes Factor method is applied to determine the risk factors that affect the prolonged hospital stay of patients postoperatively. Then the Naive Bayes Classification is applied to predict postoperative morbidity, which leads the patient to be treated for more than 4 days. Data were obtained from medical history of patients at the FundaCardio Foundation in Venezuela during the period 2010–2014. Based on the results of the Bayes factor calculation, risk factors that influenced the length of stay diagnosis in the hospital include elderly, blood transfusion of more than 2 units of PRCB (Packet of Red Cell Blood), New York Heart Association (NYHA) heart failure classification, extubation process, duration of surgery and length of stay at the ICU. By using the Naive Bayes Classification, it was found that these risk factors lead the patient to be hospitalized for more than 4 days. The results of this classification can be a consideration for patients in determining the cost of the coverage and the premium of an insurance product.