TY - JOUR
T1 - Intelligent diagnostic system for cerebrovascular diseases based on a Bayesian network with information gain
AU - Sun, Yan
AU - Bai, Ying
AU - Ding, Shuxue
AU - Tang, Yi Yuan
AU - Cui, Yifen
AU - Wang, Yan
PY - 2013/11
Y1 - 2013/11
N2 - In this paper, we present an intelligent system for analyzing the probabilistic dependencies that valuate the relationships of risk factors of cerebrovascular diseases (CVDs). We demonstrate the process used by the system to diagnose CVDs. To construct the system, we select age, gender, hypertension, diabetes mellitus, coronary heart disease, and hyperlipemia as risk factors of CVDs, which are based on the advice of experienced CVD doctors. The associations of CVDs with these risk factors are analyzed. To diagnose CVDs based on these risk factors objectively, we propose a novel system model based on a Bayesian network (BN) and information gain. By training the model using standard datasets, we obtain a diagnosis system that can automatically generate a diagnosis result when a group of data incorporating the risk factors is inputted. Finally, we test and evaluate the system using standard datasets and compare the results with those of support vector machine analysis. We also present the evaluation results from three experienced CVD doctors, who confirm that the diagnosis results of the system are beneficial to the realistic diagnosis and prediction of CVDs.
AB - In this paper, we present an intelligent system for analyzing the probabilistic dependencies that valuate the relationships of risk factors of cerebrovascular diseases (CVDs). We demonstrate the process used by the system to diagnose CVDs. To construct the system, we select age, gender, hypertension, diabetes mellitus, coronary heart disease, and hyperlipemia as risk factors of CVDs, which are based on the advice of experienced CVD doctors. The associations of CVDs with these risk factors are analyzed. To diagnose CVDs based on these risk factors objectively, we propose a novel system model based on a Bayesian network (BN) and information gain. By training the model using standard datasets, we obtain a diagnosis system that can automatically generate a diagnosis result when a group of data incorporating the risk factors is inputted. Finally, we test and evaluate the system using standard datasets and compare the results with those of support vector machine analysis. We also present the evaluation results from three experienced CVD doctors, who confirm that the diagnosis results of the system are beneficial to the realistic diagnosis and prediction of CVDs.
KW - Bayesian network (BN)
KW - Cerebrovascular diseases (CVDs)
KW - Information gain
KW - Risk factor
UR - http://www.scopus.com/inward/record.url?scp=84885773472&partnerID=8YFLogxK
M3 - Article
AN - SCOPUS:84885773472
VL - 9
SP - 4545
EP - 4554
JO - International Journal of Innovative Computing, Information and Control
JF - International Journal of Innovative Computing, Information and Control
SN - 1349-4198
IS - 11
ER -