Table 3.
Parameter estimates of the complex model with 12 estimated parameters (Nominal) and the reduced models with five sensitive parameters selected by FIM and LASSO methods ( and , respectively). Parameter 95% confidence interval width was estimated for the complex and reduced models. Goodness of fit was measured by . All values are given as mean standard deviation across subjects.
| Parameters | Parameter estimates | 95% confidence interval width | ||||
|---|---|---|---|---|---|---|
| (Nominal) | ||||||
| 428 ± 125 | 279a,b ± 27 | 401 ± 163 | 546 ± 128 | 62c,d ± 19 | 236c ± 122 | |
| 4860 ± 2205 | — | 8055a ± 1371 | 7274 ± 806 | — | 3623c ± 1418 | |
| 87.9 ± 55.3 | 16.2a,b ± 11.9 | 36.4a ± 30.5 | 123.0 ± 49.8 | 18.2c,d ± 6.0 | 68.7c ± 36.9 | |
| 0.263 ± 0.027 | 0.242b ± 0.024 | 0.226a ± 0.027 | 0.034 ± 0.010 | 0.020c,d ± 0.0710 | 0.033 ± 0.010 | |
| 0.099 ± 0.063 | 0.069 ± 0.013 | — | 0.134 ± 0.036 | 0.030c ± 0.015 | — | |
| 0.675 ± 0.280 | 0.918a ± 0.077 | — | 0.544 ± 0.145 | 0.138c ± 0.069 | — | |
| 1.733 ± 1.118 | — | 2.200 ± 1.256 | 3.086 ± 0.495 | — | 1.736c ± 0.534 | |
| 27.2 ± 16.2 | — | — | 48.7 ± 2.4 | — | — | |
| 0.433 ± 0.381 | — | — | 0.932 ± 0.075 | — | — | |
| 0.439 ± 0.325 | — | — | 0.877 ± 0.086 | — | — | |
| 1.690 ± 1.213 | — | — | 4.659 ± 0.144 | — | — | |
| 2.170 ± 1.736 | — | — | 4.758 ± 0.114 | — | — | |
| 82.26 ± 7.67 | 82.65 ± 7.57 | 81.87 ± 7.56 | — | — | — | |
The parameter estimate in the reduced model (with five estimated parameters selected by FIM and LASSO methods) was significantly different (p ≤ 0.05) from that of the complex model (with 12 estimated parameters – nominal values).
The parameter estimate in the FIM-selected parameters was significantly different (p ≤ 0.05) from that of the LASSO-selected parameters.
The parameter 95% confidence interval in the reduced model (from FIM or LASSO methods) was significantly narrower (p ≤ 0.05) than that of the complex model.
The parameter 95% confidence interval from the FIM method was significantly narrower (p ≤ 0.05) than that from the LASSO method.