Table 2:

Parameters of fits of ID and DD models to Mtb (Erdman) CFU and CEQ data in lesions of macaques.

		1 – 21 days		21 – 28 days		8 – 77 days
Model	${\delta }_D$ (day−1)	$r_1$ (day−1)	${\delta }_1$ (day−1)	$r_2$ (day−1)	${\delta }_2$ (day−1)	$r_3$ (day−1)	${\delta }_3$ (day−1)
DD model min early rates	0.000	0.422	0.000	0.989	0.782	0.079	0.134
DD model high early rates	0.000	0.915	0.492	0.915	0.708	0.079	0.134
DD model min early rates	0.036	0.422	0.000	1.066	0.859	0.810	0.865
DD model high early rates	0.036	0.998	0.575	0.998	0.791	0.810	0.865
ID model min early rates	0.000	0.422	0.000	0.401	0.194	0.006	0.062
ID model max early rates	0.000	0.487	0.065	0.207	0.000	0.006	0.062
ID model min early rates	0.036	0.458	0.036	0.437	0.230	0.042	0.098
ID model max early rates	0.036	0.523	0.101	0.243	0.036	0.042	0.098

We digitized the data on CFU and CEQ dynamics from Lin et al.¹³ and fitted alternative models to these data (see Figure 3). Because CEQs at 21 days post-infection were not shown in original publication, we considered extreme assumptions to generate a range of feasible possibilities of estimated parameters during early infection. Minimum early rates are obtained by assuming Z = 1 on day 21 of infection, and maximum early rates are obtained by setting $Z$ at 21 days post-infection equal to $Z$ at 28 days post-infection. In some cases the latter assumption led to replication rates significantly in excess of 1 day⁻¹. In these cases, we set $r\approx 1/\text{day}$. It turned out that setting $r=1/\text{day}$ in some cases led to very small differences in rate between weeks 1–3 and week 4 of infection. In these cases, we fit the data with $r$ kept constant during the first 4 weeks of infection.