Table 3.

Prediction performance of machine-learning algorithms: 6-month survival

Cancer type	Area under ROC curve (95% CI)
	EAR only	ECO only	EAR+ECO
Genitourinary	0.81 (0.77 to 0.85)	0.82 (0.78 to 0.86)	0.88 (0.85 to 0.91)*,†
Colorectal	0.84 (0.80 to 0.88)	0.85 (0.81 to 0.89)	0.88 (0.84 to 0.91)*,†
Lung	0.71 (0.67 to 0.76)	0.73 (0.69 to 0.77)*	0.77 (0.73 to 0.82)*,†
Breast	no deaths in the period
Haematological	0.73 (0.68 to 0.79)	0.74 (0.69 to 79)	0.76 (0.71 to 0.81)
Upper gastrointestinal	0.74 (0.69 to 0.78)	0.64 (0.60 to 69)	0.84 (0.80 to 0.87)†
Skin	0.84 (0.77 to 0.90)	0.85 (0.79 to 91)	0.91 (0.86 to 0.96)*,†
Head and neck	0.66 (0.61 to 0.71)	0.70 (0.64 to 75)	0.77 (0.72 to 0.82)*,†
Gynaecological	0.97 (0.94 to 0.99)	0.99 (0.98 to 1)*	1 (0.99 to 1)*
CNS	0.89 (0.85 to 0.94)	0.84 (0.78 to 0.90)	0.82 (0.77 to 0.88)
Unknown primary	0.92 (0.89 to 0.95)	0.79 (0.75 to 0.84)	0.90 (0.87 to 0.93)*,†

*Significantly greater than EAR only.

†Significantly greater than ECO only.

CNS, central nervous system; EAR, electronic administrative records; ECO, Evaluation of Cancer Outcomes; ROC, receiver operating characteristic.