January 1, 1998Medical Decision Making
Predicting Mortality after Coronary Artery Bypass Surgery
Key result
An artificial neural network model performed similarly to a logistic regression model in predicting in-hospital mortality after CABG surgery (AUC 0.78 vs 0.77; p > 0.10).
Why the study?
Does an artificial neural network model improve the prediction of in-hospital mortality after CABG surgery compared to a logistic regression model?
Population
15,608 patients undergoing coronary artery bypass graft surgery in Ontario, Canada.
Comparison
Artificial neural network model for predicting… vs Logistic regression model
Design
Cohort
Follow-up
in-hospital
Authors
Discussion
Loading...
Member takes
Implication
Similar performance supports simpler logistic regression for CABG mortality prediction; leaves open value of neural networks in larger cohorts.
Study Design
Type
Observational (n=15,608)
Structured PICO
Does an artificial neural network model improve the prediction of in-hospital mortality after CABG surgery compared to a logistic regression model?
P
Population
15,608 patients undergoing CABG surgery in Ontario, Canada, between 1991 and 1993, used to train and validate predictive models for in-hospital mortality.
E
Exposure
Artificial neural network model for predicting mortality risk
C
Comparator
Logistic regression model
O
Outcome
In-hospital mortalityhard clinical
Main Result
Absolute Event Rate: 0.78% vs 0.77%
p-value: p=> 0.10
Artificial neural networks and logistic regression models perform similarly in predicting in-hospital mortality after CABG surgery.
Cite This Study
Tu et al. (1998) conducted an observational in Coronary artery bypass graft (CABG) surgery (n=15,608). Artificial neural network model vs. Logistic regression model was evaluated on In-hospital mortality prediction (Area under the receiver operating characteristic curve) (p=> 0.10). An artificial neural network model performed similarly to a logistic regression model in predicting in-hospital mortality after CABG surgery (AUC 0.78 vs 0.77; p > 0.10).