Biomedical Engineering Reference
In-Depth Information
trial, a maximum of N = 36 patients will be treated in cohorts of size 1, starting at
(c;q) = (0.20, 0.10). The admissibility limits are T = 0.15 and E = 0.50, and the
probability cut-offs are pT E = p T = .95.
Priors were established in three steps. In step 1, for each (c;q); we elicited prior
means of the probability of dissolving the clot immediately with the bolus, p 0 (c;q;),
within 60 minutes, F E ( 2 j c;q;), or within 120 minutes, F E (1 j c;q;). We also
elicited the probability of toxicity, T (y E ;c;q;); if the clot was dissolved instanta-
neously (y E = 0), if it was dissolved within the 120 minute infusion (y E = 1), or if
it was not dissolved during the infusion (y E > 1). The elicited values are given in
Table 12.2.
TABLE 12.2: Elicited Prior Mean Probabilities for Each (c;q) Studied in
the IA tPA Trial.
c = 0.20 c = 0.30 c = 0.40 c = 0.50
q = 0.10
p 0 (c;q;)
0.10
0.15
0.15
0.25
F E ( 2 j c;q;)
0.25
0.30
0.45
0.50
F E (1 j c;q;)
0.35
0.45
0.60
0.70
T (0;c;q;)
0.02
0.03
0.03
0.03
T (1;c;q;)
0.04
0.06
0.08
0.12
q = 0.20
p 0 (c;q;)
0.15
0.20
0.25
0.30
F E ( 2 j c;q;)
0.40
0.45
0.50
0.60
F E (1 j c;q;)
0.50
0.60
0.70
0.80
T (0;c;q;)
0.02
0.03
0.03
0.03
T (1;c;q;)
0.04
0.06
0.08
0.12
Ef T (:50;q; 1(Y E > 1);)g = 0.15
For q = 0.10 or 0.20,
For the second step, the elicited values were treated like the true state of
nature and used to simulate 1,000 pseudo-samples, each of size 400 with 50
patients given each (c;q) combination. Starting with a noninformative pseudo-
prior on in which the logarithm of each entry followed an N(0; 20 2 ) distribu-
Search WWH ::




Custom Search