ID Exposure Route # of Doses Agent Strain Dose Units Host type Μodel LD50/ID50 Optimized parameters Response type Reference
281 injected in eyelids 6.00 ATCC 19660 CFU Swiss webster mice (5day old) exponential 8.13E+03 k = 8.52E-05 death
Hazlett, L. D., Rosen, D. D., & Berk, R. S. (1978). Age-Related Susceptibility to Pseudomonas aeruginosa Ocular Infections in Mice. Infection and Immunity, 20, 1.
281,282 (pooled) injected in eyelids 12.00 ATCC 19660 CFU Swiss webster mice (5day old) exponential 6.61E+03 k = 1.05E-04 death
Hazlett, L. D., Rosen, D. D., & Berk, R. S. (1978). Age-Related Susceptibility to Pseudomonas aeruginosa Ocular Infections in Mice. Infection and Immunity, 20, 1.
282 injected in eyelids 6.00 ATCC 19660 CFU Swiss webster mice (5day old) exponential 4.98E+03 k = 1.39E-04 death(after day 21)
Hazlett, L. D., Rosen, D. D., & Berk, R. S. (1978). Age-Related Susceptibility to Pseudomonas aeruginosa Ocular Infections in Mice. Infection and Immunity, 20, 1.
283 injected in eyelids 6.00 ATCC 19660 CFU Swiss webster mice(10day old) beta-Poisson 1.93E+04 a = 6.73E-01 N50 = 1.93E+04 death(after day 1)
Hazlett, L. D., Rosen, D. D., & Berk, R. S. (1978). Age-Related Susceptibility to Pseudomonas aeruginosa Ocular Infections in Mice. Infection and Immunity, 20, 1.
283,284 injected in eyelids 12.00 ATCC 19660 CFU Swiss webster mice(10day old) beta-Poisson 1.48E+04 a = 6.01E-01 N50 = 1.48E+04 death
Hazlett, L. D., Rosen, D. D., & Berk, R. S. (1978). Age-Related Susceptibility to Pseudomonas aeruginosa Ocular Infections in Mice. Infection and Immunity, 20, 1.
284 injected in eyelids 6.00 ATCC 19660 CFU Swiss webster mice(10day old) beta-Poisson 1.13E+04 a = 5.49E-01 N50 = 1.13E+04 death(after day 2-21)
Hazlett, L. D., Rosen, D. D., & Berk, R. S. (1978). Age-Related Susceptibility to Pseudomonas aeruginosa Ocular Infections in Mice. Infection and Immunity, 20, 1.
286 contact lens 5.00 ATCC 19660 CFU Swiss webster mice exponential 7.88E+06 k = 8.8E-08 infection (Keratitis)
Hazlett, L. D., Rosen, D. D., & Berk, R. S. (1978). Age-Related Susceptibility to Pseudomonas aeruginosa Ocular Infections in Mice. Infection and Immunity, 20, 1.
297 contact lens 5.00 CFU white rabbit beta-Poisson 6.57E+03 a = 3.55E-01 N50 = 6.57E+03 corneal ulceration
Lawin-Brüssel, C. A., Refojo, M. F., Leong, F. L., Hanninen, L. ., & Kenyon, K. R. (1993). Effect of Pseudomonas aeruginosa concentration in experimental contact lens-related microbial keratitis. Cornea, 12, 1.
297 & 298 contact lens 10.00 CFU white rabbit beta-Poisson 1.85E+04 a = 1.9E-01 N50 = 1.85E+04 corneal ulceration
Lawin-Brüssel, C. A., Refojo, M. F., Leong, F. L., Hanninen, L. ., & Kenyon, K. R. (1993). Effect of Pseudomonas aeruginosa concentration in experimental contact lens-related microbial keratitis. Cornea, 12, 1.
298 contact lens 5.00 CFU white rabbit beta-Poisson 1.52E+05 a = 1.09E-01 N50 = 1.52E+05 Severe stromal ulceration
Lawin-Brüssel, C. A., Refojo, M. F., Leong, F. L., Hanninen, L. ., & Kenyon, K. R. (1993). Effect of Pseudomonas aeruginosa concentration in experimental contact lens-related microbial keratitis. Cornea, 12, 1.
Ojielo2003 intratracheal 7.00 PA01 CFU C57BL/6 mice exponential k = 3.22E-7 death
Ojielo, C. I., Cooke, K. ., Mancuso, P. ., Standiford, T. J., Olkiewicz, K. M., Clouthier, S. ., … Moore, B. B. (2003). Defective Phagocytosis and Clearance of <i>Pseudomonas aeruginosa</i> in the Lung Following Bone Marrow Transplantation. The Journal of Immunology, 171(8), 4416-4424. Retrieved from https://journals.aai.org/jimmunol/article/171/8/4416/35775/Defective-Phagocytosis-and-Clearance-of (Original work published)
Exposure Route:
injected in eyelids
# of Doses:
6.00
Agent Strain:
ATCC 19660
Dose Units:
CFU
Host type:
Swiss webster mice (5day old)
Μodel:
exponential
LD50/ID50:
8.13E+03
Optimized parameters: k = 8.52E-05
Response type:
death

mice (5day old) /  Pseudomonas aeruginosa 
Dose Dead Survived Total
2.5 0 13 13
25 0 14 14
250 0 12 12
2500 4 13 17
25000 13 2 15
250000 17 0 17

 

Goodness of fit and model selection
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential

0.802

0.0885 5 3.84 
0.766
11.1 
0.977
Beta Poisson

0.713

4 9.49 
0.95
Exponential is preferred to beta-Poisson; cannot reject good fit for exponential.

 

Optimized k parameter for the exponential model, from 10000 bootstrap iterations
Parameter MLE estimate Percentiles
0.5% 2.5% 5% 95% 97.5% 99.5%
k 8.52E-05 4.21E-05 4.99E-05 5.50E-05 1.53E-04 1.57E-04 2.05E-04
ID50/LD50/ETC* 8.13E+03 3.38E+03 4.40E+03 4.54E+03 1.26E+04 1.39E+04 1.65E+04
*Not a parameter of the exponential model; however, it facilitates comparison with other models.

 

Parameter histogram for exponential model (uncertainty of the parameter)

Exponential model plot, with confidence bounds around optimized model

Highest quality
Exposure Route:
injected in eyelids
# of Doses:
12.00
Agent Strain:
ATCC 19660
Dose Units:
CFU
Host type:
Swiss webster mice (5day old)
Μodel:
exponential
LD50/ID50:
6.61E+03
Optimized parameters: k = 1.05E-04
Response type:
death

mice (5day old, day 1-21 pooled) /  Pseudomonas aeruginosa 

Dose Dead Survived Total
2.5 0 13 13
2.5 0 13 13
25 0 14 14
25 0 14 14
250 0 12 12
250 0 12 12
2500 4 13 17
2500 4 13 17
25000 13 2 15
25000 15 0 15
250000 17 0 17
250000 17 0 17

 

Goodness of fit and model selection
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential

4.36 

 -0.000549

  3.84 
1
19.7
0.958
Beta Poisson

4.36

10 18.3 
0.93
Exponential is preferred to beta-Poisson; cannot reject good fit for exponential.

 

Optimized k parameter for the exponential model, from 500 bootstrap iterations
Parameter MLE estimate Percentiles
0.5% 2.5% 5% 95% 97.5% 99.5%
k 1.05E-04 6.84E-05 7.49E-05 7.84E-05 1.48E-04 1.57E-04 1.73E-04
ID50/LD50/ETC* 6.61E+03 4.01E+03 4.40E+03 4.68E+03 8.84E+03 9.26E+03 1.01E+04
*Not a parameter of the exponential model; however, it facilitates comparison with other models.

 

 

 

Parameter histogram for exponential model (uncertainty of the parameter)

Exponential model plot, with confidence bounds around optimized model

Exposure Route:
injected in eyelids
# of Doses:
6.00
Agent Strain:
ATCC 19660
Dose Units:
CFU
Host type:
Swiss webster mice (5day old)
Μodel:
exponential
LD50/ID50:
4.98E+03
Optimized parameters: k = 1.39E-04
Response type:
death(after day 21)

mice (5day old) /  Pseudomonas aeruginosa 
Dose Dead Survived Total
2.5 0 13 13
25 0 14 14
250 0 12 12
2500 4 13 17
25000 15 0 15
250000 17 0 17

 

Goodness of fit and model selection
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential

2.17

-0.000196 

5 3.84 
1
11.1 
0.825
Beta Poisson

2.17

4 9.49 
0.704
Exponential is preferred to beta-Poisson; cannot reject good fit for exponential.

 

Optimized k parameter for the exponential model, from 10000 bootstrap iterations
Parameter MLE estimate Percentiles
0.5% 2.5% 5% 95% 97.5% 99.5%
k 1.39E-04 8.87E-05 9.90E-05 9.90E-05 2.05E-04 2.36E-04 2.73E-04
ID50/LD50/ETC* 4.98E+03 2.54E+03 2.94E+03 3.38E+03 7.00E+03 7.00E+03 7.82E+03
*Not a parameter of the exponential model; however, it facilitates comparison with other models.

 

Parameter histogram for exponential model (uncertainty of the parameter)

Exponential model plot, with confidence bounds around optimized model

Exposure Route:
injected in eyelids
# of Doses:
6.00
Agent Strain:
ATCC 19660
Dose Units:
CFU
Host type:
Swiss webster mice(10day old)
Μodel:
beta-Poisson
LD50/ID50:
1.93E+04
Optimized parameters: a = 6.73E-01 N50 = 1.93E+04
Response type:
death(after day 1)

mice (10day old, day1) /  Pseudomonas aeruginosa 
Dose Dead Survived Total
2.5 0 10 10
25 0 16 16
250 0 15 15
2500 0 13 13
25000 12 4 16
250000 14 3 17

 

Goodness of fit and model selection
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential

21.1

13.8 5 3.84 
0.000207
11.1 
0.000783
Beta Poisson

7.32

4 9.49 
0.12
Beta-Poisson fits better than exponential; cannot reject good fit for beta-Poisson.

 

Optimized parameters for the beta-Poisson model, from 10000 bootstrap iterations
Parameter MLE estimate Percentiles
0.5% 2.5% 5% 95% 97.5% 99.5%
α 6.73E-01 3.02E-01 3.62E-01 3.80E-01 1.55E+00 6.30E+03 1.06E+04
N50 1.93E+04 9.62E+03 1.12E+04 1.18E+04 3.42E+04 4.01E+04 5.18E+04

 

 

 

Parameter scatter plot for beta Poisson model ellipses signify the 0.9, 0.95 and 0.99 confidence of the parameters.

beta Poisson model plot, with confidence bounds around optimized model

Exposure Route:
injected in eyelids
# of Doses:
12.00
Agent Strain:
ATCC 19660
Dose Units:
CFU
Host type:
Swiss webster mice(10day old)
Μodel:
beta-Poisson
LD50/ID50:
1.48E+04
Optimized parameters: a = 6.01E-01 N50 = 1.48E+04
Response type:
death

mice (10day old, day1-21) /  Pseudomonas aeruginosa 
Dose Dead Survived Total
2.5 0 10 10
2.5 0 10 10
25 0 16 16
25 0 16 16
250 0 15 15
250 0 15 15
2500 0 13 13
2500 2 11 13
25000 12 4 16
25000 13 3 16
250000 14 3 17
250000 14 3 17

 

Goodness of fit and model selection
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential

48.6 

36.3 11 3.84 
 1.65e-09
19.7
1.09e-06
Beta Poisson

12.3

10 18.3 
0.265
Beta-Poisson fits better than exponential; cannot reject good fit for beta-Poisson.

 

Optimized parameters for the beta-Poisson model, from 500 bootstrap iterations
Parameter MLE estimate Percentiles
0.5% 2.5% 5% 95% 97.5% 99.5%
α 6.01E-01 3.11E-01 3.69E-01 3.97E-01 1.04E+00 1.20E+00 2.01E+00
N50 1.48E+04 8.01E+03 9.07E+03 1.00E+04 2.29E+04 2.55E+04 2.83E+04

 

 

 

Parameter scatter plot for beta Poisson model ellipses signify the 0.9, 0.95 and 0.99 confidence of the parameters.

beta Poisson model plot, with confidence bounds around optimized model

Exposure Route:
injected in eyelids
# of Doses:
6.00
Agent Strain:
ATCC 19660
Dose Units:
CFU
Host type:
Swiss webster mice(10day old)
Μodel:
beta-Poisson
LD50/ID50:
1.13E+04
Optimized parameters: a = 5.49E-01 N50 = 1.13E+04
Response type:
death(after day 2-21)

mice (10day old, day2-21) /  Pseudomonas aeruginosa 
Dose Dead Survived Total
2.5 0 10 10
25 0 16 16
250 0 15 15
2500 2 11 13
25000 13 3 16
250000 14 3 17

 

Goodness of fit and model selection
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential

27.3

23.1 5 3.84 
 1.54e-06
11.1
5.01e-05
Beta Poisson

4.19

4 9.49 
0.381
Beta-Poisson fits better than exponential; cannot reject good fit for beta-Poisson.

 

Optimized parameters for the beta-Poisson model, from 10000 bootstrap iterations
Parameter MLE estimate Percentiles
0.5% 2.5% 5% 95% 97.5% 99.5%
α 5.49E-01 2.38E-01 2.87E-01 3.17E-01 1.37E+00 1.13E+01 1.15E+04
N50 1.13E+04 4.53E+03 5.70E+03 6.34E+03 2.17E+04 2.46E+04 3.39E+04

 

 

 

Parameter scatter plot for beta Poisson model ellipses signify the 0.9, 0.95 and 0.99 confidence of the parameters.

beta Poisson model plot, with confidence bounds around optimized model

Exposure Route:
contact lens
# of Doses:
5.00
Agent Strain:
ATCC 19660
Dose Units:
CFU
Host type:
Swiss webster mice
Μodel:
exponential
LD50/ID50:
7.88E+06
Optimized parameters: k = 8.8E-08
Response type:
infection (Keratitis)

Swiss webster mice /  Pseudomonas aeruginosa 
Dose Infection (keratitis)

Not Infection (keratitis)

Total
1E+04 0 8 8
1E+05 1 9 10
1E+06 1 7 8
1E+07 5 5 10
1E+08 10 0 10

 

Goodness of fit and model selection
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential

3.6

0.134 4 3.84 
0.715
9.49 
0.464
Beta Poisson

3.46

3 7.81 
0.326
Exponential is preferred to beta-Poisson; cannot reject good fit for exponential.

 

Optimized k parameter for the exponential model, from 10000 bootstrap iterations
Parameter MLE estimate Percentiles
0.5% 2.5% 5% 95% 97.5% 99.5%
k 8.8E-08 3.31E-08 3.89E-08 4.53E-08 1.58E-07 1.94E-07 2.47E-07
ID50/LD50/ETC* 7.88E+06 2.81E+06 3.57E+06 4.38E+06 1.53E+07 1.78E+07 2.09E+07
*Not a parameter of the exponential model; however, it facilitates comparison with other models.

 

Parameter histogram for exponential model (uncertainty of the parameter)

Exponential model plot, with confidence bounds around optimized model

Exposure Route:
contact lens
# of Doses:
5.00
Agent Strain:
Dose Units:
CFU
Host type:
white rabbit
Μodel:
beta-Poisson
LD50/ID50:
6.57E+03
Optimized parameters: a = 3.55E-01 N50 = 6.57E+03
Response type:
corneal ulceration

White rabbit /  Pseudomonas aeruginosa 
Dose Corneal Ulceration

Not Corneal Ulceration

Total
63.2 0 5 5
2220 2 3 5
13200 3 2 5
78000 3 2 5
462000 5 0 5

 

Goodness of fit and model selection
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential

9.41

7.02 4 3.84 
0.00807 
9.49 
0.0516
Beta Poisson

2.39

3 7.81 
0.495
Beta-Poisson fits better than exponential; cannot reject good fit for beta-Poisson.

 

Optimized parameters for the beta-Poisson model, from 10000 bootstrap iterations
Parameter MLE estimate Percentiles
0.5% 2.5% 5% 95% 97.5% 99.5%
α 3.55E-01 1.53E-01 1.76E-01 1.91E-01 4.52E+00 1.73E+07 1.35E+11
N50 6.57E+03 7.71E+02 1.44E+03 1.70E+03 2.55E+04 4.30E+04 8.32E+04

 

Parameter histogram for exponential model (uncertainty of the parameter)

Exponential model plot, with confidence bounds around optimized model

Highest quality
Exposure Route:
contact lens
# of Doses:
10.00
Agent Strain:
Dose Units:
CFU
Host type:
white rabbit
Μodel:
beta-Poisson
LD50/ID50:
1.85E+04
Optimized parameters: a = 1.9E-01 N50 = 1.85E+04
Response type:
corneal ulceration

Pooled data of White rabbit /  Pseudomonas aeruginosa 
Dose Corneal Ulceration

Not Corneal Ulceration

Total
63.2 0 5 5
63.2 0 5 5
2220 2 3 5
2220 2 3 5
13200 3 2 5
13200 1 4 5
78000 3 2 5
78000 1 4 5
462000 5 0 5
462000 4 1 5

 

Goodness of fit and model selection
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential

30.9

20.8 9 3.84 
5.11e-06 
16.9 
0.000312
Beta Poisson

10.1

8 15.5 
0.26
Beta-Poisson fits better than exponential; cannot reject good fit for beta-Poisson.

 

Optimized parameters for the beta-Poisson model, from 10000 bootstrap iterations
Parameter MLE estimate Percentiles
0.5% 2.5% 5% 95% 97.5% 99.5%
α 1.9E-01 9.63E-02 1.15E-01 1.24E-01 3.86E-01 5.51E-01 2.50E+00
N50 1.85E+04 3.18E+03 4.73E+03 6.01E+03 7.05E+04 8.98E+04 1.97E+05

 

Parameter histogram for exponential model (uncertainty of the parameter)

Exponential model plot, with confidence bounds around optimized model

Exposure Route:
contact lens
# of Doses:
5.00
Agent Strain:
Dose Units:
CFU
Host type:
white rabbit
Μodel:
beta-Poisson
LD50/ID50:
1.52E+05
Optimized parameters: a = 1.09E-01 N50 = 1.52E+05
Response type:
Severe stromal ulceration

White rabbit /  Pseudomonas aeruginosa 
Dose Severe Stromal Ulceration

Not Corneal Ulceration

Total
63.2 0 5 5
2220 2 3 5
13200 1 4 5
78000 1 4 5
462000 4 1 5

 

Goodness of fit and model selection
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential

13.2

8.73 4 3.84 
0.00313 
9.49 
0.0105
Beta Poisson

4.43

3 7.81 
0.219
Beta-Poisson fits better than exponential; cannot reject good fit for beta-Poisson.

 

Optimized parameters for the beta-Poisson model, from 10000 bootstrap iterations
Parameter MLE estimate Percentiles
0.5% 2.5% 5% 95% 97.5% 99.5%
α 1.09E-01 4.05E-02 5.23E-02 5.72E-02 9.05E+06 4.82E+07 2.53E+12
N50 1.52E+05 4.66E+03 1.13E+04 1.55E+04 1.55E+07 6.25E+07 1.96E+09

 

Parameter histogram for exponential model (uncertainty of the parameter)

Exponential model plot, with confidence bounds around optimized model

Exposure Route:
intratracheal
# of Doses:
7.00
Agent Strain:
PA01
Dose Units:
CFU
Host type:
C57BL/6 mice
Μodel:
exponential
LD50/ID50:
Optimized parameters: k = 3.22E-7
Response type:
death

Table 1: Goodness of fit
Model Deviance Δ Degrees 
of freedom
χ20.95,1 
p-value
χ20.95,m-k 
p-value
Exponential 16.88 0.001 6 3.84 
0.98
12.6
0.009
Beta Poisson 16.88 5 11.1
0.004
Neither model provides a good fit to the data

 

Table 2: Multi-hit dose response statistics
Model Deviance Δ DF χ20.95,df χ20.95,1 Good fit? Parameters LD50
Multi-hit 1.09 15.69 5 11.1 3.84 Yes k = 4.12 × 10−6 kmin=11 2,588,047

 

Figure 1. The multi-hit dose response model with 95% and 99% confidence bands
Figure 1. The multi-hit dose response model with 95% and 99% confidence bands

 

 

Figure 2. Histogram of the k parameter estimates after bootstrapping
Figure 2. Histogram of the k parameter estimates after bootstrapping

 

References