Table 5 Performance of DL (Yolo) risk model with respect to the different number of training samples for the training and validation cohorts.
From: Radiomics-guided deep neural networks stratify lung adenocarcinoma prognosis from CT scans
Test | ||||
|---|---|---|---|---|
Number of samples | Hazard ratio (95% CI) | p-value | C-index | |
Whole (n = 432) | 7.6277 (2.2870–25.4399) | 0.0027 | 0.7593 | |
300 | 6.4256 (1.9081–21.6386) | 0.0071 | 0.7950 | |
200 | 4.0644 (1.2211–13.5284) | 0.0478 | 0.6368 | |
100 | 2.6913 (0.8064–8.9821) | 0.1927 | 0.5988 | |
Validation | ||||
Number of samples | Hazard ratio (95% CI) | p-value | C-index | |
Whole (n = 432) | 6.5362 (2.1773–19.6213) | 0.0022 | 0.7696 | |
300 | 3.9521 (1.3726–11.3794) | 0.0228 | 0.7724 | |
200 | 2.4826 (0.8542–7.2149) | 0.1620 | 0.5135 | |
100 | 2.2463 (0.8442–7.1475) | 0.1686 | 0.5391 | |
