Table 4.
The performance metrics of the neural network model using Adam using the default threshold (50%).
| Sociodemographic groups | Recall | Specificity | Accuracy | AUCa | Differenceb | P value | |||||||||
| Sex | 21.31 | <.001 | |||||||||||||
|
|
Male | 57.14 | 97.69 | 97.50 | 97.49 |
|
|
||||||||
|
|
Female | 77.78 | 97.02 | 96.87 | 95.83 |
|
|
||||||||
| Marital status | 10.40 | <.001 | |||||||||||||
|
|
Never been married | 76.47 | 96.98 | 96.85 | 97.81 |
|
|
||||||||
|
|
Other groups of marital status | 66.67 | 97.58 | 97.39 | 95.71 |
|
|
||||||||
| Working condition | 9.71 | <.001 | |||||||||||||
|
|
Working ≥35 hours | 77.78 | 97.73 | 97.67 | 97.69 |
|
|
||||||||
|
|
Other groups of working condition | 68.75 | 97.05 | 96.82 | 95.82 |
|
|
||||||||
| Race | 46.57 | <.001 | |||||||||||||
|
|
White | 79.31 | 97.07 | 96.95 | 97.66 |
|
|
||||||||
|
|
Black | 33.33 | 97.66 | 97.36 | 94.33 |
|
|
||||||||
| Income | 15.38 | <.001 | |||||||||||||
|
|
An income of <US $20,000 | 80.00 | 94.68 | 94.54 | 97.43 |
|
|
||||||||
|
|
Other groups of income | 67.74 | 97.80 | 97.63 | 96.42 |
|
|
||||||||
aAUC: area under the curve.
bDifference between recall and specificity values.