Machine learning-based radiomics score improves prognostic prediction accuracy of stage II/III gastric cancer: A multi-cohort study

Table 1 Baseline data of the discovery and the Cancer Imaging Archive cohorts, n (%)

	Discovery cohort	TCIA cohort	P value
	(n = 141)	(n = 40)
Gender			0.002
Male	79 (56.0)	33 (82.5)
Female	62 (44.0)	7 (17.5)
Age (yr)	56.3 (12.1)	65.0 (9.6)	< 0.001
ASA			NA
1	70 (49.6)	Not reported
2	40 (28.4)
3	31 (22.0)
CEA (ng/mL)			NA
≤ 5	109 (77.3)	Not reported
> 5	32 (22.7)
Operation			< 0.001
Proximal	18 (12.8)	17 (42.5)
Distal	84 (59.6)	11 (27.5)
Total	39 (27.7)	12 (30.0)
Tumor size	4.3 (1.2)	Not reported	NA
Nerve or vascular invasion			NA
Yes	54 (38.3)	Not reported
No	87 (61.7)
SMI	46.8 (14.5)	44.8 (13.4)	0.009
SMD	33.2 (6.8)	34.6 (7.2)	0.032
Differentiation			NA
Good	13 (9.2)	Not reported
Moderate	63 (44.7)
Poor	65 (46.1)
T stage			0.097
2-3	66 (46.8)	23 (57.5)
4	75 (53.2)	17 (42.5)
N stage			< 0.001
0-1	33 (23.4)	16 (40.0)
> 1	108 (76.6)	24 (60.0)
TD			NA
Yes	29 (20.6)	Not reported
No	112 (79.4)
TNM stage
II	65 (46.1)	12 (30.0)	< 0.001
III	76 (53.9)	28 (70.0)
Chemotherapy			NA
Yes	90 (63.8)	Not reported
No	51 (36.2)

SMI: Skeletal muscle index; SMD: Skeletal muscle density; TD: Tumor deposition; NA: Not available; TCIA: The Cancer Imaging Archive; TNM: Tumor, node, and metastasis; ASA: American Association of Anesthesiologists score; CEA: Carcinoembryonic antigen.