Table 1.
ML information criteria for choosing mean model.
| ML IC | PDX | Form of mean model | ||||
|---|---|---|---|---|---|---|
| Cubic + interaction | Cubic | Quadratic + interaction | Quadratic | Straight + interaction | ||
| AIC | PH80 | 199.3 | 194.1 | 193 | 191 | 189.2* |
| PH87 | 124.5 | 121 | 119.1 | 115.8* | 117.5 | |
| PH77 | 388.7 | 383.2 | 381.6 | 377* | 383.1 | |
| PH95 | 164.8 | 160.8 | 159 | 157* | 173.4 | |
| PH39 | 336.5 | 333.9* | 334.2 | 339 | 339.6 | |
| BIC | PH80 | 253.8 | 239 | 234.8 | 223.1 | 218.1* |
| PH87 | 176 | 163.4 | 158.5 | 146.1 | 144.8* | |
| PH77 | 449.7 | 433.4 | 428.3 | 412.9* | 415.4 | |
| PH95 | 217 | 203.8 | 198.9 | 187.6* | 201 | |
| PH39 | 389.7 | 377.7 | 374.8 | 370.3 | 367.7* | |
Lower values are better for both AIC and BIC, and the lowest per PDX model is indicated by *. AIC favors more complex models, while BIC includes a penalty for the number of parameters estimated so tends to favor more simple models with fewer parameters. Bold font indicates the chosen mean model.
ML Maximum likelihood, AIC Akaike information criterion, BIC Bayesian information criterion, PDX Patient derived xenograft.