TABLE 2.

Accuracy of visual acuity predictions.

Algorithm learner	1 month		3 months		6 months
Validation set	MAE	RMSE	MAE	RMSE	MAE	RMSE
Full model	0.057 ± 0.008	0.080 ± 0.011	0.070 ± 0.013	0.099 ± 0.021	0.072 ± 0.011	0.098 ± 0.019
Simplified model Ⅰ	0.057 ± 0.007	0.080 ± 0.010	0.071 ± 0.014	0.099 ± 0.023	0.073 ± 0.011	0.099 ± 0.018
Simplified model Ⅱ	0.058 ± 0.008	0.080 ± 0.010	0.072 ± 0.012	0.099 ± 0.019	0.075 ± 0.010	0.102 ± 0.016
Simplified model Ⅲ	0.077 ± 0.007	0.100 ± 0.010	0.079 ± 0.012	0.110 ± 0.022	0.080 ± 0.010	0.107 ± 0.014
XEC set	MAE	RMSE	MAE	RMSE	MAE	RMSE
Simplified model Ⅰ	0.074 (0.060–0.089)	0.096 (0.075–0.114)	0.074 (0.051–0.100)	0.103 (0.066–0.137)	0.092 (0.058–0.127)	0.121 (0.075–0.157)
Simplified model Ⅱ	0.074 (0.060–0.114)	0.096 (0.078–0.114)	0.073 (0.051–0.098)	0.101 (0.066–0.134)	0.094 (0.062–0.133)	0.123 (0.081–0.160)
Simplified model Ⅲ	0.083 (0.066–0.101)	0.110 (0.085–0.137)	0.089 (0.065–0.118)	0.119 (0.083–0.155)	0.098 (0.064–0.138)	0.127 (0.084–0.165)

MAE, mean absolute error; RMSE, root mean square error; XEC, Xiamen Eye Center. Accuracy (VA in logMAR) of VA prediction at 1, 3, and 6 months after laser treatment compared with the ground truth. The results were stratified according to the follow-up period and the points inputted into the algorithms; this table shows only the predictive effect of the baseline data. All VA predictions in the validation set are shown with the standard deviation (in logMAR); all VA predictions in the XEC set are shown with the 95% confidence interval.