Predicting in-hospital rupture of type A aortic dissection using Random Forest.

Authors:
Mohammad A Zafar, MBBS
Mohammad A Zafar, MBBS
Aortic Institute at Yale-New Haven Hospital
Research Director
New Haven, CT | United States

J Thorac Dis 2019 Nov;11(11):4634-4646

Department of Cardiovascular Surgery, State Key Laboratory of Cardiovascular Disease, Fuwai Hospital, National Center for Cardiovascular Disease, Chinese Academy of Medical Sciences and Peking Union Medical College, Beijing 100037, China.

Background: This study is to establish prediction tools for in-hospital rupture of type A aortic dissection (TAAD) patients, to better guide emergency surgical triage and patient counselling.

Methods: We retrospectively evaluated 1,133 consecutive patients with TAAD from January 2010 to December 2016. The study population was divided into training and testing datasets in a 70:30 ratio for further analysis using Random Forest.

Results: The Random Forest classification model was developed with the training dataset and 16 variables were confirmed as 'important': age, BMI, gender, syncope, lower limb numbness/pain, acute phase of the TAAD, BP >160 mmHg at admission; acute liver dysfunction, WBC >15×10/L, aortic size, aortic height index (AHI), periaortic hematoma, pleural effusion, brachiocephalic artery involvement, renal artery involvement, and hemopericardium. Validation of the model showed good discrimination with an AUC, sensitivity, specificity, positive predictive value and negative predictive value of 0.994, 1.000, 0.987, 0.998 and 1.000, respectively, in the training dataset, and 0.752, 0.990, 0.514, 0.945 and 0.857, respectively, in the testing dataset.

Conclusions: An easy-to-use tool to predict in-hospital rupture for TAAD patients was developed and validated (http://47.107.228.109/). Periaortic hematoma is the strongest predictor. Simple clinical information such as syncope can be very useful in in-hospital rupture risk stratification.

Download full-text PDF

Source
http://dx.doi.org/10.21037/jtd.2019.10.82DOI Listing
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC6940214PMC
November 2019

Publication Analysis

Top Keywords

in-hospital rupture
16
aortic dissection
8
type aortic
8
artery involvement
8
taad patients
8
random forest
8
training dataset
8
rupture type
8
periaortic hematoma
8
developed training
4
brachiocephalic artery
4
dataset variables
4
testing datasetconclusions
4
confirmed 'important'
4
0945 0857
4
age bmi
4
predictive 0994
4
'important' age
4
0857 testing
4
effusion brachiocephalic
4

Similar Publications