Pathologist-trained machine learning classifiers developed to quantitate celiac disease features differentiate endoscopic biopsies according to modified marsh score and dietary intervention response

Table 1 Composition of the study training and test sets

Characteristic	Total Study (N = 116)	Training Set (N = 10)	Test Set (N = 106)
Age category	Count (%)	Count (%)	Count (%)
≤ 21 years	75 (65)	6 (60)	69 (65)
> 21 years	41 (35)	4 (40)	37 (35)
Sex
Male	58 (50)	6 (60)	52 (49)
Female	58 (50)	4 (40)	54 (51)
Documented biopsy location
Duodenal bulb	4 (3)	0 (0)	4 (4)
Duodenum, NOS	98 (84)	10 (100)	88 (76)
Small intestine, NOS	14 (12)	0 (0)	14 (13)
Modified Marsh classification
Type 0^a	42 (36)	2 (20)	40 (38)
Type 1	10 (9)	1 (10)	9 (8)
Type II	2 (2)	0 (0)	2 (2)
Type IIIa	23 (20)	2 (20)	21 (20)
Type IIIb	37 (32)	5 (50)	32 (30)
Type IIIc	2 (2)	0 (0)	2 (2)

ISSN: 1746-1596