Table 8

Performance of each algorithm in the validation cohort in men and women (including patients with imputed data)

StatisticWomen: mean (95% CI)Men: mean (95% CI)
Blood cancer
 D statistic1.639 (1.578 to 1.699)1.726 (1.673 to 1.78)
 R2 (%)39.1 (37.3 to 40.8)41.6 (40.1 to 43.1)
 ROC statistic0.803 (0.796 to 0.811)0.8 (0.793 to 0.807)
Breast cancer
 D statistic1.088 (1.058 to 1.119)NA
 R2 (%)22 (21.1 to 23)NA
 ROC statistic0.761 (0.758 to 0.765)NA
Bowel cancer
 D statistic1.974 (1.922 to 2.027)2.139 (2.091 to 2.188)
 R2 (%)48.2 (46.9 to 49.5)52.2 (51.1 to 53.3)
 ROC statistic0.847 (0.842 to 0.852)0.862 (0.858 to 0.866)
Gastro-oesophageal cancer
 D statistic2.277 (2.181 to 2.372)2.241 (2.174 to 2.308)
 R2 (%)55.3 (53.2 to 57.4)54.5 (53 to 56)
 ROC statistic0.873 (0.864 to 0.881)0.868 (0.862 to 0.874)
Lung cancer
 D statistic2.742 (2.687 to 2.797)2.797 (2.75 to 2.844)
 R2 (%)64.2 (63.3 to 65.1)65.1 (64.4 to 65.9)
 ROC statistic0.905 (0.901 to 0.91)0.911 (0.908 to 0.914)
Oral cancer
 D statistic1.817 (1.67 to 1.964)1.881 (1.771 to 1.991)
 R2 (%)44.1 (40.1 to 48.1)45.8 (42.9 to 48.7)
 ROC statistic0.795 (0.775 to 0.814)0.808 (0.794 to 0.823)
Ovarian cancer
 D statistic1.311 (1.237 to 1.385)NA
 R2 (%)29.1 (26.8 to 31.4)NA
 ROC statistic0.769 (0.76 to 0.778)NA
Pancreas cancer
 D statistic2.235 (2.126 to 2.345)2.225 (2.119 to 2.33)
 R2 (%)54.4 (52 to 56.8)54.2 (51.8 to 56.5)
 ROC statistic0.865 (0.855 to 0.875)0.857 (0.847 to 0.867)
Prostate cancer
 D statisticNA2.252 (2.219 to 2.285)
 R2 (%)NA54.8 (54 to 55.5)
 ROC statisticNA0.895 (0.893 to 0.897)
Renal tract cancer
 D statistic2.005 (1.923 to 2.086)2.234 (2.181 to 2.287)
 R2 (%)49 (46.9 to 51)54.4 (53.2 to 55.5)
 ROC statistic0.851 (0.843 to 0.859)0.863 (0.858 to 0.867)
Uterine cancer
 D statistic1.758 (1.677 to 1.839)NA
 R2 (%)42.5 (40.2 to 44.7)NA
 ROC statistic0.828 (0.819 to 0.837)NA
  • Notes on understanding validation statistics.

  • Discrimination is the ability of the risk prediction model to differentiate between patients who experience a admission event during the study and those who do not. This measure is quantified by calculating the area under the receiver operating characteristic curve (ROC) statistic; where a value of 1 represents perfect discrimination.

  • The D statistic is also a measure of discrimination which is specific to censored survival data. As with the ROC, higher values indicate better discrimination.

  • R2 measures explained variation and higher values indicate more variation is explained.

  • NA, not applicable.