Table 2 Multivariable odds ratios for the machine-learned features

From: Predicting lymph node metastasis from primary tumor histology and clinicopathologic factors in colorectal cancer using deep learning

 

Temporal validation

External validation 1a

External validation 1b

Covariate

OR [95% CI]

p

OR [95% CI]

p

OR [95% CI]

p

Age

 < 60

1.00 (reference)

n/a

1.00 (reference)

n/a

1.00 (reference)

n/a

 60–69

0.48 [0.29, 0.79]

0.004

0.58 [0.35, 0.96]

0.033

0.51 [0.35, 0.74]

< 0.001

 70–79

0.51 [0.32, 0.83]

0.007

0.39 [0.23, 0.64]

< 0.001

0.41 [0.27, 0.61]

< 0.001

 > 80

0.51 [0.31, 0.85]

0.009

0.38 [0.21, 0.67]

< 0.001

0.35 [0.22, 0.57]

< 0.001

Sex

 Male

1.00 (reference)

n/a

1.00 (reference)

n/a

1.00 (reference)

n/a

 Female

1.52 [1.09, 2.13]

0.014

1.01 [0.70, 1.47]

0.953

0.91 [0.68, 1.23]

0.545

Tumor Grade

 G1

1.00 (reference)

1.00 (reference)

1.00 (reference)

n/a

1.00 (reference)

n/a

 G2

0.72 [0.45, 1.15]

0.166

1.09 [0.75, 1.57]

0.65

0.78 [0.58, 1.04]

0.094

 G3

1.18 [0.70, 1.99]

0.534

1.66 [0.81, 3.41]

0.169

1.29 [0.68, 2.43]

0.44

T-Category

 T2

n/a

n/a

n/a

n/a

1.00 (reference)

n/a

 T3

1.00 (reference)

n/a

1.00 (reference)

n/a

1.56 [1.14, 2.13]

0.005

 T4

1.24 [0.81, 1.88]

0.319

1.04 [0.60, 1.81]

0.884

1.87 [1.11, 3.16]

0.018

Lymphatic Invasion

 L0

1.00 (reference)

1.00 (reference)

1.00 (reference)

n/a

1.00 (reference)

n/a

 L1 + 

1.76 [1.16, 2.68]

0.008

5.48 [3.36, 8.96]

< 0.001

4.06 [2.76, 5.96]

< 0.001

Venous Invasion

 V0

1.00 (reference)

1.00 (reference)

1.00 (reference)

n/a

1.00 (reference)

n/a

 V1 + 

1.85 [1.12, 3.04]

0.016

1.58 [0.83, 2.99]

0.163

1.53 [0.88, 2.66]

0.132

Machine-learned Features

 1

1.88 [1.33, 2.68]

< 0.001

1.31 [0.47, 3.64]

0.611

1.54 [0.67, 3.54]

0.31

 2

0.50 [0.20, 1.29]

0.154

0.59 [0.32, 1.08]

0.086

0.66 [0.42, 1.05]

0.078

 3

2.18 [1.48, 3.21]

< 0.001

1.64 [1.18, 2.27]

0.003

1.45 [1.13, 1.86]

0.003

 4

0.80 [0.60, 1.07]

0.13

0.62 [0.44, 0.88]

0.007

0.63 [0.46, 0.85]

0.003

 5

0.87 [0.58, 1.32]

0.516

1.04 [0.58, 1.86]

0.907

0.77 [0.49, 1.21]

0.264

  1. Odds ratios (OR) from multivariable logistic regression model baseline clinicopathologic variables (age, sex, tumor grade, T-category, lymphatic invasion, venous invasion) and 5 machine-learned features.
Back to article page