Table 7.

Comparison of model performance between non-AI methods and AI models. AUC, area under the curve; MAE, mean absolute error

Reference	Performance of the non-AI method	Performance of the AI model
Constantinou et al. 2015 [46]	AUC score range from 0.665 to 0.717	AUC score is 0.78
Stern et al. 2016 [34]	Error between 0.65 and 0.72 years	Best MAE is 0.36 ± 0.3 years
Spampinato et al. 2017 [35]	Error is 30% higher than the AI model	MAE is 0.79 years
Milosevic et al. 2019 [16]	Accuracy ranges from 0.71 to 0.95	Accuracy is 0.9687 ± 0.0096
Turan et al. 2019 [17]	Accuracy ranges from 0.807 to 0.901	Accuracy is 0.95
Peleg et al. 2020 [18]	Accuracy is 0.845 for Australian and 0.865 for African American	Accuracy is 0.863 for European American, 0.82 for Israeli, and 0.816 for African American population
Peña-Solorzano et al. 2020 [19]	Error in the range of 1 to 11 mm	MAE is 2 mm
Vila-Blanco et al. 2020 [42]	Best median and mean error are − 0.02 ± 0.71 and − 0.04 years respectively. Best MAE is 0.488 years	Median and mean error are − 0.01 ± 0.8 and − 0.04 years respectively. MAE is 0.72 years