Proefschrift

Treatment Schedule t CR Volume Variance T − No Treatment > 3000 5 . 657 × 10 7 1 . 338 × 10 7 1 . 200 × 10 7 SOC 203 4 . 396 × 10 7 1 . 424 × 10 7 4 . 373 × 10 7 20% SOC 2388 4 . 978 × 10 7 9 . 429 × 10 6 2 . 101 × 10 7 LFHA w/ Induction 203 4 . 399 × 10 7 1 . 410 × 10 7 4 . 439 × 10 7 LFHA w/o Induction 366 4 . 405 × 10 7 1 . 357 × 10 7 4 . 338 × 10 7 HFLA w/ Induction 1319 4 . 808 × 10 7 9 . 464 × 10 6 2 . 444 × 10 7 HFLA w/o Induction 1461 4 . 776 × 10 7 8 . 816 × 10 6 2 . 565 × 10 7 Volume 203 4 . 396 × 10 7 1 . 422 × 10 7 4 . 371 × 10 7 Variance 1950 4 . 907 × 10 7 7 . 996 × 10 6 2 . 175 × 10 7 T − 2416 4 . 966 × 10 7 1 . 274 × 10 7 1 . 708 × 10 7 Table A.2: Time to Competitive Release and Comparative Objectives for the “Responder” Patient. Treatment Schedule t CR Volume Variance T − No Treatment 458 5 . 822 × 10 7 1 . 018 × 10 7 2 . 708 × 10 7 SOC 60 4 . 922 × 10 7 9 . 685 × 10 6 4 . 910 × 10 7 20% SOC 410 5 . 437 × 10 7 8 . 421 × 10 6 3 . 456 × 10 7 LFHA w/ Induction 60 4 . 922 × 10 7 9 . 685 × 10 6 4 . 910 × 10 7 LFHA w/o Induction 60 4 . 927 × 10 7 9 . 442 × 10 6 4 . 894 × 10 7 HFLA w/ Induction 624 5 . 309 × 10 7 8 . 343 × 10 6 3 . 720 × 10 7 HFLA w/o Induction 391 5 . 274 × 10 7 7 . 877 × 10 6 3 . 810 × 10 7 Volume 60 4 . 922 × 10 7 9 . 685 × 10 6 4 . 910 × 10 7 Variance 458 5 . 372 × 10 7 7 . 671 × 10 6 3 . 544 × 10 7 T − 2402 5 . 408 × 10 7 9 . 759 × 10 6 3 . 137 × 10 7 Table A.3: Time to Competitive Release and Comparative Objectives for the “Non- Responder” Patient. To quantify the robustness of the bang-bang treatment schedules to jitter, the 100 minimum objective values are compared to the total range of objective values for each objective. We chose 100 as it is large enough to show significant perturbation from the minimum objective value yet small enough to allow visual representation of the associated treatment schedules. The percent of the total range of possible objective values covered by the minimum 100 objective values is used as a metric to measure sensitivity. A visual representation of this comparison is shown in Figure A.1. The ratio of the range of the smallest 100 objective values to the total range of objective values is defined as Ω and used as a measure of robustness. If the range of the smallest 100 objective values spans only a small percentage of the total possible range of objective values, the bang-bang treatment schedule is said to be robust, as small changes to the treatment schedule do not increase the objective value a significant amount. On the other hand, if the minimum 100 objective values spans a large percentage of the total possible range of objective values, the 127