Fig. 5

Comparison of logistic [74] and probit [17] in Schultheiss et al. [75] showed that for some datasets these two models may differ by less than 1% over their entire range. For this dataset as shown above, the two models differ by no more than ± 1.6% from their average, up to 60 Gy in 3–4 fractions, but diverge at doses above that where the data is sparse. Abbreviations IU Indiana University, AE adverse event (brachial plexopathy)