Single-time-point dosimetry using model selection and nonlinear mixed-effects modelling: a proof of concept

EJNMMI Physics

Table 1 The goodness of fits and Akaike weights for the investigated functions in the NLME model

Equation number	Function name	K	Akaike weight (%)	Jackknife Akaike weights (% mean (SD); % median [min, max])^a	RD (% median [min, max])^b	RMSE of the RD (%)^b
1	\(f_{3a}\)	7	0.08	–	− 3.94 [− 4.39, − 2.05]	3.66
2	\(f_{3b}\)	7	0.09	–	− 3.79 [− 5.99, − 2.19]	4.06
3	\(f_{3c}\)	7	0.09	–	− 3.89 [− 4.61, − 2.16]	3.69
4	\(f_{3d}\)	7	0.11	–	− 3.64 [− 6.17, − 2.28]	4.04
5	\(f_{4a}\)	9	0.77	–	− 0.78 [− 1.72, 0.56]	0.97
6	\(f_{4b}\)	9	7.35	10 (7); 11 [0,21]	− 0.69 [− 1.91, 0.44]	1.09
7	\(f_{4c}\) ^c	9	–	–	–	–
8	\(f_{4d}\)	9	47.26	45 (6); 43 [38,56]	–	–
9	\(f_{4e}\)	9	44.25	45 (7); 43 [37,57]	− 0.49[− 0.64,− 0.04]	0.49
10	\(f_{5a}\) ^c	11	–	–	–	–
11	\(f_{5b}\) ^c	11	–	–	–	–

The total number of biokinetic data N used in this analysis is 40; the number of parameters of the NLME model for the corresponding SOE function is given in column K
^aThe Jackknife analysis was based on functions with Akaike weights > 1%, i.e. \(f_{4b}\), \(f_{4d}\), and \(f_{4e}\)
^bFunction \(f_{4d}\) was used as the reference for the calculation of the RDs and RMSEs of the all-time-point TIAs
^cFunctions \(f_{4c}\), \(f_{5a}\), and \(f_{5b}\) did not pass the goodness-of-fit test because CV > 50%; therefore, Akaike weights, RDs, and RMSEs were not calculated for these functions