Table 1.

Parameters of the 4 alternative models fit to Datasets 1–4 (excluding data with $\mathrm{CTL}={\mathbf{10}}^{\mathbf{8}}\mathrm{cell}/\mathrm{mL}$) and metrics of relative quality of the model fits. Estimated parameters: $\alpha$ is a dimensionless inverse of the effective fraction; r is given in units of per day; h is in cell/mL; $g_1$ is in per day; $g_2$ is in cell/mL. The parameter k had different units in different models: per OT1 cell/mL per day, per day, per OT1 (cell/mL)${}^n$ per day, and per OT1 cell/mL per day for MA (Equation (3)), Sat (Equation (4)), Power (Equation (5)), and SiGMA (Equation (6)), respectively; n is a dimensionless parameter. The parameter estimates and 95% confidence intervals for the best-fit SiGMA model were: $\alpha =2.71$(2.5–2.9), $g_0=0.12$(0.036–0.2)/day, $g_1=0.64$(0.55–0.73)/day, $g_2=6.72$(4.14–17.57) $\times {10}^3\mathrm{cell}/\mathrm{mL}$, and $k=3.3$(3.15–3.4)$\times {10}^{-7}\mathrm{cell}/\mathrm{mL}/\mathrm{day}$; the model fits are shown in Figure 2. The best-fit model (with the highest w) is highlighted in blue.

Datasets 1–4 ($\mathit{E}\le {10}^7$ cell/mL): n = 431
Model	$\alpha$	r, 1/day	k	h	n	$g_0$	$g_1$	$g_2$	SSR	AIC	$\Delta$	w
MA	2.77	0.576	$3.79\times {10}^{-7}$						164	814	160	0
Sat	2.81	0.744	6	$5.34\times {10}^6$					118.5	677	23	0
Power	2.79	0.744	$2.23\times {10}^{-4}$		0.606				116	668	14	0
SiGMA	2.71		$3.29\times {10}^{-7}$			0.12	0.64	6715	112	654	0	1