Application of machine learning in predicting survival outcomes involving real-world data: a scoping review

Table 3 Descriptive statistics of AUC by ML algorithms

ML category	Number of models^d	Mean (STD)	Median	Min	Max	IQR
Random survival forests	13	0.8084	0.821	0.64	0.9503	0.73–0.85
Boosted tree methods^a	5	0.7876	0.78	0.722	0.853	0.73–0.853
Artificial neural networks	11	0.7999	0.802	0.721	0.926	0.747–0.8208
Support Vector Machine	3	0.7633	0.8	0.64	0.85	0.72–0.825
Regularization^b	6	0.7164	0.7095	0.6	0.801	0.709–0.7546
Other algorithms^c	4	0.7899	0.7695	0.7287	0.8917	0.7447–0.8147

Abbreviations: ML Machine learning, NNs Neural networks, RF Random forest, DT Decision tree, SVM Support vector machine, STD Standard deviation, IQR The interquartile range.
^aIt includes adaboost, gradient boosting, gradient descent boosting, boosting, XGBoost
^bIt includes Lasso (L1 regularization), ridge regression (L2 regularization), and elastic-net algorithms
^cIt includes: naives bayes, KNN or MLP
^dthe total number of studies may differ from than total number of included studies, because some studies used more than 1 ML algorithms and also models with no AUC reported were excluded

ISSN: 1471-2288