Table 2

Performance comparison set results

Unadjusted ME in optimisation dataset ±SDME in performance comparison dataset ±SDMAE in performance comparison dataset ±SDMedAE in performance comparison dataset
OKULIX−0.329±0.4710.018±0.4780.357±0.3180.280
ML0.211±0.4700.019±0.4700.351±0.3130.274
% improvement//1.7%2.1%
  • The first column shows the ME ±SD in the optimisation dataset, where the ME was calculated from the unadjusted predictions from OKULIX and ML. The second column shows the ME ±SD in the performance comparison dataset, after subtracting the corresponding ME in the optimisation dataset from the original prediction from OKULIX and ML. The per cent improvement was calculated as Embedded Image.

  • MAE, mean absolute error; ME, mean error; MedAE, median absolute error; ML, machine learning.