Table 4 Characteristics of the final learning data.

From: Convolutional neural network for classification of two-dimensional array images generated from clinical information may support diagnosis of rheumatoid arthritis

 

RA

nonRA

 

(n = 252)

(n = 785)

RF (IU/mL, median, range)

68 (0–2265)

5 (0–810)

P < 0.0001

ACPA (U/mL, median, range)

(0–3519)

0 (0–145)

P < 0.0001

ESR (mm/hour, median, range)

36 (5–111)

10 (5–67)

P < 0.0001

Symptoms onset to visit

(days, median, quartile range)

60 (30–120)

60 (30–360)

P = 0.0098

CRP (mg/dL, median, range)

0.63 (0–14.8)

0.05 (0–5.6)

P < 0.0001

ANA (n times, median, range)

0 (0–640)

0 (0–2560)

P = 0.0026

MMP-3 (ng/mL, median, range)

116 (16–706)

38.3 (10–155)

P < 0.0001

WBC (/uL, median, range)

7200 (2700–12200)

5700 (2100–12800)

P < 0.0001

gender (female/male)

150/102

585/200

P < 0.0001

skin abnormality (positive/negative)

0/252

25/760

P = 0.09

body temperature (°C, range)

35.8–37.4

35.8–37.8

P = 0.08

body trunk pain (positive/negative)

36/216

168/617

P = 0.014

patient-VAS (0–100, median, range)

35 (0–90)

10 (0–90)

P < 0.0001

doctor-VAS (0–100, median, range)

10 (0–80)

1 (0–60)

P < 0.0001

  1. *Abbreviation, RA = rheumatoid arthritis, RF = rheumatoid factor, ACPA = anti-citrullinated protein/peptide antibody, ESR = erythrocyte sedimentation rate, CRP = C-related protein, ANA = anti-nuclear antibody, MMP-3 = matrixmetalloproteinase-3, WBC = white blood cell count and VAS = visual analog scale.