Blood Glucose Level Prediction for Diabetics Based on Nutrition and Insulin Administration Logs Using Personalized Mathematical Models

Table 5

Results for the full time period with extended model parameter set and Nelder–Mead downhill simplex method (values in mmol/l).

Extended parameter set and Nelder–Mead algorithm test (full data with restart)	Average error	RMSE	EGA − A + B (acceptable)
Slow GA with 3 parameters	2.94	4.09	88.6%
Slow GA with 6 parameters	2.11	2.96	92.2%
Slow GA with 9 parameters	2.16	3.05	92.1%
Slow GA with 6 parameters + Nelder–Mead method	1.98	2.76	92.5%