Skip to main content

Table 4 Logistic regression analyses for the prediction of local recurrence

From: Combined PET/CT-perfusion in patients with head and neck cancers might predict failure after radio-chemotherapy: a proof of concept study

 

Coefficient

SE

β estimate

95 % CI

p

c statistics

(a) Model for the probability of recurrence according to BVpre

BVpre

5.636

1.696

280.418

−0.685 – 6.162

0.08

0.867

(b) Model for the probability of recurrence according to SUVmax

SUVmaxpre

0.036

4.603

1.036

−1.426 – 19.166

0.770

0.600

(c) Model for the probability of recurrence according to BVpre and SUVmax

BVpre

1.992

0.144

7.329

−1.568 – 2.126

0.003

 

SUVmaxpre

−3.048

1.447

0.047

−7.379 – -0.384

0.040

0.900

  1. SE Standard error
  2. CI confidence interval
  3. C-statistics: accuracy of chosen statistical model
  4. β estimate: resulting from a regression analysis hat have been standardized