Prediction of preoperative acute ischemic stroke risk in patients with acute type A aortic dissection based on neural network
Journal Title: Journal of Air Force Medical University - Year 2023, Vol 44, Issue 8
Abstract
Objective To construct the risk assessment model of preoperative acute ischemic stroke ( AIS) for acute type A aortic dissection ( ATAAD ) by using neural network, and to identify the risk factors associated with preoperative AIS in ATAAD patients. Methods A retrospective analysis was conducted on 261 ATAAD patients diagnosed by computed tomography angiography ( CTA) and receiving brain magnetic resonance diffusion-weighted imaging scanning from January 2015 to September 2018. Patients were divided into AIS ( + ) group (n = 79) and AIS ( - ) group (n = 182) according to whether they had preoperative AIS. With clinical and imaging features as input layer, a preoperative AIS risk prediction model was constructed based on neural network. Accuracy, sensitivity, specificity and area under the curve (AUC) were used to evaluate the prediction performance of the model,and the model was compared with the preoperative AIS prediction models constructed by traditional univariate and multivariate logistic regression method and by the least absolute shrinkage and selection operator ( LASSO) regression. In addition, LASSO regression was used to evaluate the contribution of each input feature to the neural network features. Results The true lumen diameter and true lumen diameter ratio of ascending aorta in AIS ( + ) group were significantly lower than those in AIS ( - ) group ( P < 0. 01). The incidence of intimal flap plaque (P < 0. 01), common carotid artery dissection (P < 0. 01), innominate artery or common carotid artery originating from the false lumen ( P < 0. 01), reduction of the density of unilateral internal carotid artery compared with the contralateral side (P < 0. 01), subclavian artery dissection (P < 0. 01), and reduction of the density of unilateral vertebral artery compared with the contralateral side (P < 0. 05) was higher in the AIS ( + ) group than that in the AIS ( - ) group. Among the three prediction models, the model based on neural network method had the best prediction performance, and its accuracy, sensitivity, specificity, and AUC were 0. 886 ± 0. 043, 0. 900 ± 0. 073, 0. 881 ± 0. 054 and 0. 925 ± 0. 039, respectively. LASSO regression was used to fit the risk factors extracted by the neural network. It was found that age, the true lumen diameter ratio of ascending aorta, common carotid artery dissection and reduction of the density of unilateral internal carotid artery compared with the contralateral side played a key role in the fitting process. Conclusion Specific carotid and aortic CTA imaging features are associated with preoperative AIS in ATAAD patients. Neural network has a good performance in preoperative AIS risk prediction in ATAAD patients.
Authors and Affiliations
XUE Ruijia, ZHENG Minwen, JIN Zhenxiao, DUAN Weixun, LIU Wen, WANG Jinfeng, ZHAO Hongliang
Keywords
Related Articles
- EP ID EP724931
- DOI -
- Views 25
- Downloads 0
Network construction of key modulatory miRNA and downstream target genes in the progression of osteoporosis under simulated microgravity through high-throughput sequencing
Objective To screen the differential expression profiles of miRNAs in normal and weightless osteoporosis cell samples by using high-throughput sequencing technology, and to analyze the biological information of the diff...
A multicenter, double-blind, randomized placebo-controlled phase Ⅱ / Ⅲ clinical trial of meplazumab in the treatment of COVID-19
Objective Meplazumab is a humanized anti-CD147 IgG2 monoclonal antibody, which can inhibit the virus from invading human host cells by blocking the interaction between CD147 and the spike protein of severe acute respir...
Prevention and treatment of oxidative stress injury of vitiligo based on the TCM theory of “preventive treatment of disease”
The TCM theory of “preventive treatment of disease” can play a role in the prevention and treatment of many diseases. We found that the three states in the TCM theory of “preventive treatment of disease”, namely healthy...
Genome-wide association analysis of POCD susceptibility in elderly patients after cardiac surgery with cardiopulmonary bypass
Objective To screen the susceptibility gene loci for cognitive dysfunction after cardiac surgery with cardiopulmonary bypass (CPB) in elderly patients. Methods A total of 120 patients scheduled for cardiac surgery with...
Investigation and analysis of oral health of officers and soldiers in a unit of airborne troops
Objective To investigate and analyze the oral health status of paratroopers in a unit of airborne troops, and to provide basis for targeted diagnosis, treatment and prevention of oral diseases in airborne troops. Method...