Table 2.
Subgroup | Number of included studies | Sensitivity (95% CI) | Specificity (95% CI) | PLRa | NLRb | DORc | AUCd | |
Value of meta-analysis in all included studies | 6 | 0.87 (0.72-0.94) | 0.86 (0.77-0.92) | 6.2 (3.8-10.1) | 0.15 (0.07-0.34) | 40 (15-112) | 0.92 (0.90-0.94) | |
Methodological quality of included studiese |
|
|
|
|
|
|
|
|
|
High quality | 5 | 0.89 (0.75-0.96) | 0.88 (0.83-0.92) | 7.7 (5.6-10.6) | 0.12 (0.05-0.28) | 64 (32-129) | 0.94 (0.91-0.95) |
|
Low quality | 1 | Null | Null | Null | Null | Null | Null |
Total number of included patientse |
|
|
|
|
|
|
|
|
|
≤100 | 4 | 0.90 (0.73-0.97) | 0.88 (0.81-0.93) | 7.6 (5.3-10.9) | 0.11 (0.04-0.32) | 68 (29-158) | 0.94 (0.91-0.95) |
|
<100 | 2 | Null | Null | Null | Null | Null | Null |
Format of study |
|
|
|
|
|
|
|
|
|
Retrospective | 4 | 0.90 (0.73-0.97) | 0.88 (0.81-0.93) | 7.6 (5.3-10.9) | 0.11 (0.04-0.32) | 68 (29-158) | 0.94 (0.91-0.95) |
|
Prospective | 2 | Null | Null | Null | Null | Null | Null |
Published year |
|
|
|
|
|
|
|
|
|
After 2010 | 4 | 0.80 (0.64-0.90) | 0.86 (0.73-0.93) | 5.6 (2.8-11.3) | 0.24 (0.13-0.45) | 23 (8-72) | 0.90 (0.87-0.92) |
|
Before 2010 | 2 | Null | Null | Null | Null | Null | Null |
Type of AIf |
|
|
|
|
|
|
|
|
|
Neural network–based | 4 | 0.78 (0.64-0.87) | 0.87 (0.74-0.94) | 6.0 (2.7-13.0) | 0.26 (0.15-0.44) | 23 (7-73) | 0.89 (0.86-0.91) |
|
SVMg-based | 2 | Null | Null | Null | Null | Null | Null |
Type of endoscopic image |
|
|
|
|
|
|
|
|
|
WLIh | 5 | 0.86 (0.67-0.95) | 0.86 (0.75-0.92) | 6.1 (3.4-10.9) | 0.16 (0.06-0.42) | 37 (11-124) | 0.92 (0.89-0.94) |
|
LCIi | 1 | Null | Null | Null | Null | Null | Null |
Classifying performance between Helicobacter pylori–positive vs H pylori–naïve patients | 2 | 0.82 (0.74-0.89) | 0.85 (0.81-0.89) | 3.5 (0.8-14.3) | 0.27 (0.05-1.41) | 13 (0.8-229) | Null |
aPLR: positive likelihood ratio.
bNLR: negative likelihood ratio.
cDOR: diagnostic odds ratio.
dAUC: area under the curve.
eThese modifiers were significant in the meta-regression analysis.
fAI: artificial intelligence.
gSVM: support vector machine.
hWLI: white-light imaging.
iLCI: linked color imaging.