Table 2 Estimated coefficients (uncorrected and corrected for optimism) of pre-treatment and ∆image biomarkers models fitted to the entire dataset.
From: Delta-radiomics features during radiotherapy improve the prediction of late xerostomia
Model | β | odds ratio (95% CI) | p-value | ||
|---|---|---|---|---|---|
Uncorrected | Corrected | ||||
Pre-treatment reference model | intercept | −3.794 | −3.385* | ||
Xerbaseline | 2.531 | 2.280* | 12.56 (3.39–46.54) | <0.001 | |
Parotid gland dose | 0.099 | 0.089* | 1.1 (1.03–1.18) | 0.005 | |
∆feature model 1 | intercept | −3.139 | −2.515 | ||
Xerbaseline | 2.533 | 2.074 | 12.59 (3.13–50.73) | <0.001 | |
∆PG-surfacew3 (cm2) | −0.568 | −0.465 | 0.57 (0.41–0.79) | 0.001 | |
∆feature model 2 | intercept | −4.515 | −3.305 | ||
Xerbaseline | 2.591 | 1.936 | 13.35 (3.13–56.95) | <0.001 | |
Parotid gland dose | 0.072 | 0.054 | 1.07 (0.77–1.51) | 0.074 | |
∆PG-surfacew3 (cm2) | −0.481 | −0.360 | 0.62 (0.57–0.67) | 0.005 | |