Determination of Risk Factors for Predicting Pulmonary Embolism Using Bayesian Networks



Feizmanesh F¹ ; Safaei AA¹ ; Moghadam KG² Show Affiliations
Source: Koomesh Published:2018

Abstract

Introduction: Pulmonary embolism is the third leading cause of cardiovascular death after Myocardial infarction and stroke. At the same time, it is the most preventable cause of death for hospitalized patients. Importantly the diagnosis and prediction of pulmonary embolism requires flexible decision-making models, both for the presence of clinical interventions as well as for the variety of local diagnostic resources, Bayesian networks that fully meet these needs. Accordingly determining the risk factors for pulmonary embolism in hospitalized patients and presenting the model for predicting its occurrence through modeling using Bayesian networks have been proposed as a therapeutic necessity. Materials and Methods: The present research is descriptive-analytic study. The data used in the study included risk factors affecting the pulmonary embolism and the history of hospitalized patients in pulmonary section of Shariati hospital in Tehran were collected in Excel format. Bayesian prediction model in two modes (risk factors determined using the proposed scenario and risk factors according to the expert physician) is obtained using GENIE software and the accuracy of the diagnosis of pulmonary embolism was evaluated. Results: The results showed that among the risk factors of the disease, the history of thromboembolic pulmonary, history of deep vein thrombosis, body mass index above 30, recent surgery, immobilization of long-term, SLE, antiphospholipid syndrome, heart failure and pneumonia respectively, are the most important risk factors for pulmonary embolism. And the model predicts the scenario proposed has better performance. Conclusion: Such plans can facilitate the process of assessing the risk of pulmonary embolism in hospitalized patients, in order to facilitate appropriate preventive measures, and to improve preventive methods and, consequently, diagnosis and treatment programs. © 2018, Semnan University of Medical Sciences. All rights reserved.