Table 3.
Preferences for using deep learning in community-based eye disease screening.
| Attribute and level | Medical staffa | Residents | |||
|
|
ORb (95% CI) | OR (95% CI) | |||
| Diagnostic technology | |||||
|
|
Semiautomated DLc model | 2.08 (1.71, 2.52)d | 0.89 (0.68, 1.15) | ||
|
|
Fully automated DL model | 2.39 (1.97, 2.90)d | 0.24 (0.20, 0.29)d | ||
|
|
Fully manual diagnosis | Reference | Reference | ||
| Organizational form | |||||
|
|
Centralized screening | Reference | Reference | ||
|
|
Residents’ health self-examination cabin | Not significant | Not significant | ||
|
|
Opportunity screening in outpatiente | Not significant | Not significant | ||
| Missed diagnosis rate | |||||
|
|
None | Reference | Reference | ||
|
|
5% | Not significant | Not significant | ||
|
|
10% | Not significant | Not significant | ||
|
|
15% | Not significant | Not significant | ||
|
|
20% | Not significant | Not significant | ||
| Overdiagnosis rate | |||||
|
|
None | Reference | Reference | ||
|
|
5% | 0.88 (0.68, 1.15) | Not significant | ||
|
|
10% | 0.61 (0.46, 0.81)d | Not significant | ||
|
|
15% | 0.63 (0.48, 0.83)f | Not significant | ||
|
|
20% | 0.51 (0.38, 0.68)d | Not significant | ||
| Costg | |||||
|
|
Free | Reference | Reference | ||
|
|
40 CNY | 0.61 (0.46, 0.83)f | 0.75 (0.56, 1.01) | ||
|
|
80 CNY | 0.47 (0.35, 0.64)d | 0.56 (0.42, 0.74)d | ||
|
|
120 CNY | 0.39 (0.28, 0.54)d | 0.82 (0.51, 1.31) | ||
|
|
160 CNY | 0.27 (0.19, 0.38)d | 0.78 (0.46, 1.32) | ||
|
|
200 CNY | 0.21 (0.15, 0.29)d | 0.57 (0.46, 0.71)d | ||
| Screening result feedback form | |||||
|
|
Screening results | Not significant | 0.52 (0.44, 0.61)d | ||
|
|
Screening results and referral recommendations | Not significant | 0.75 (0.65, 0.87)d | ||
|
|
Screening results, referral recommendations, and oral explanation by GPh | Reference | Reference | ||
| Screening result feedback efficiency | |||||
|
|
Immediately | Reference | Reference | ||
|
|
In 2 weeks | 0.68 (0.56, 0.82)d | Not significant | ||
|
|
In 1 month | 0.58 (0.48, 0.70)d | Not significant | ||
aIn each grid, an OR value over 1 means that the health care services providers were more inclined to this level, while the value less than 1 means that they disliked this level even more.
bOR: odds ratio.
cDL: deep learning.
dP<.001.
eResidents with chronic diseases and other risk factors would be recommended by general practitioners for eye disease screening during their outpatient follow-up.
fP=.001.
gIn 2021, US $1= 6.5 CNY.
hGP: general practitioner.