Table 5.
Pooled mean estimates of highest specificity by several factors.
| Group | Studies, Na | Sample size, N | Specificity (%), range | Pooled mean specificity, mean (95% CIb) | Heterogeneity measures | Test for subgroup differences (P value) | |||||||||||||
|
|
|
|
|
|
τ2 | Q (P value) | I2 (%) |
|
|||||||||||
| Algorithms | .78 | ||||||||||||||||||
|
|
Random forest | 5 | 10,705 | 0.56-1.00 | 0.90 (0.71-1.00) | 0.0658 | 208.5 (<.001) | 98.1 |
|
||||||||||
|
|
Support vector machine | 3 | 10,554 | 0.88-1.00 | 0.96 (0.84-1.00) | 0.0325 | 189.6 (<.001) | 98.9 |
|
||||||||||
|
|
Decision tree | 3 | 10,895 | 0.77-1.00 | 0.95 (0.76-1.00) | 0.0623 | 608.1 (<.001) | 99.7 |
|
||||||||||
|
|
Multilayer perceptron | 3 | 298 | 0.73-0.91 | 0.87 (0.83-0.91) | <0.0001 | 2.3 (.33) | 11.0 |
|
||||||||||
|
|
Logistic regression | 2 | 46 | 0.73-0.77 | 0.77 (0.63-0.88) | 0.0000 | 0.1 (.71) | 0.0 |
|
||||||||||
|
|
XGBoost | 2 | 880 | 0.52-0.91 | 0.75 (0.31-1.00) | 0.1070 | 150.2 (<.001) | 99.3 |
|
||||||||||
| Aims of AIc | .11 | ||||||||||||||||||
|
|
Detectionb | 17 | 41,470 | 0.56-1.00 | 0.94 (0.65-0.99) | 1.3743 | 42,583.8 (<.001) | 100 |
|
||||||||||
|
|
Predictionb | 7 | 4085 | 0.52-0.94 | 0.77 (0.01-0.97) | 0.3083 | 361.9 (<.001) | 100 |
|
||||||||||
| Status of WDd | .62 | ||||||||||||||||||
|
|
Commercialb | 20 | 44,795 | 0.52-1.00 | 0.93 (0.61-0.99) | 1.4583 | 70,885.1 (<.001) | 100 |
|
||||||||||
|
|
Noncommercialb | 4 | 760 | 0.70-0.97 | 0.89 (0.41-0.98) | 0.0592 | 96.9 (<.001) | 97.6 |
|
||||||||||
| Data sources | .82 | ||||||||||||||||||
|
|
WD-basedb | 15 | 40,959 | 0.52-1.00 | 0.93 (0.52-0.99) | 1.5424 | 40,154.8 (<.001) | 100 |
|
||||||||||
|
|
WD-based and othersb | 9 | 4596 | 0.77-0.97 | 0.90 (0.84-0.94) | 0.0000 | 318.5 (<.001) | 97.8 |
|
||||||||||
| Reference standards | .88 | ||||||||||||||||||
|
|
STAIb,e | 7 | 148 | 0.70-0.91 | 0.83 (0.65-0.92) | 0.0294 | 10.2 (.12) | 51.9 |
|
||||||||||
|
|
CIDIf | 2 | 483 | 0.52-0.96 | 0.77 (0.27-1.00) | 0.1470 | 143.3 (<.001) | 99.3 |
|
||||||||||
| Validation methods | .50 | ||||||||||||||||||
|
|
K-foldb | 18 | 44,467 | 0.52-1.00 | 0.95 (0.53-1.00) | 1.8181 | 68,899.9 (<.001) | 100 |
|
||||||||||
|
|
Leave-one-out | 2 | 328 | 0.56-0.94 | 0.79 (0.34-1.00) | 0.1047 | 15.6 (<.001) | 93.6 |
|
||||||||||
| All studiesb | 24 | 45,555 | 0.52-1.00 | 0.92 (0.68-0.98) | 1.1844 | 75,736.0 (<.001) | 100 | N/Ag | |||||||||||
aMany studies were included more than once in each meta-analysis given that they assessed the performance of more than one algorithm.
bSpecificity was pooled using the multilevel meta-analysis method.
cAI: artificial intelligence.
dWD: wearable device.
eSTAI: State-Trait Anxiety Inventory.
fCIDI: Composite International Diagnostic Interview.
gN/A: not applicable.