Eur J Cardiothorac Surg. 2023 Apr 21:ezad160. doi: 10.1093/ejcts/ezad160. Online ahead of print.
ABSTRACT
OBJECTIVES: We aim to develop the first risk prediction models for 30-day mortality for benchmarking outcomes specifically for the Australian and New Zealand patient populations and examine whether machine learning algorithms outperform traditional statistical approaches.
METHODS: Data from the Australia New Zealand Congenital Outcomes Registry for Surgery, which contains information on every paediatric cardiac surgical encounter in Australian and New Zealand for patients aged <18 years between January 2013 to December 2021 was analysed (n = 14,343). The outcome was mortality within the 30-day period following a surgical encounter, with approximately 30% of the observations randomly selected to be used for validation of the final model. Three different ML methods were used, all of which employed 5-fold cross-validation to prevent overfitting, with model performance judged primarily by the area under the receiver operating curve (AUC).
RESULTS: Among the 14,343 30-day periods there were 188 deaths (1.3%). In the validation data, the gradient boosted tree obtained the best performance [AUC = 0.87, 95% CI = (0.82, 0.92); calibration = 0.97 95% confidence intervals = (0.72, 1.27)], outperforming penalised logistic regression and artificial neural networks [AUC of 0.82 and 0.81 respectively]. The strongest predictors of mortality in the GBT were patient weight, STAT score, age, and gender.
CONCLUSIONS: Our risk prediction model outperformed logistic regression and achieved a level of discrimination comparable to the PRAiS2 and STS-CHSD mortality risk models (both which obtained AUC = 0.86). Non-linear ML methods can be used to construct accurate clinical risk prediction tools.
PMID:37084239 | DOI:10.1093/ejcts/ezad160
