JMIR Publications

ABSTRACT

Background:

Deep learning-assisted eye diseases diagnosis technology is increasingly applied in the eye diseases screening. However, none research has suggested the prerequisites for healthcare service providers and residents to be willing to use it.

Objective:

To reveal the preferences of healthcare service providers and residents for using artificial intelligence in community-based eye disease screening, particularly the preference for accuracy.

Methods:

Two discrete choice experiments were conducted in Shanghai, China. In total 34 medical institutions with adequate AI-assisted screening experience participated. A total of 39 healthcare service providers and 318 residents were asked to answer the questionnaire and make a trade-off among the difference levels of each attribute. Conditional logit models with the stepwise selection method were used to estimate the preferences.

Results:

Healthcare service providers preferred high accuracy: Both the sensitivity and the specificity should be more than 90%, which were much higher than the FDA’s standards. However, the residents did not care about the accuracy. They preferred to have the doctors involved in the screening process, and leave the choice of the accuracy to their general practitioners. In addition, when compared with a fully manual diagnosis, AI technology was more favored by the health service providers, while the residents hated the AI technology without doctors’ supervision.

Conclusions:

Both the sensitivity and specificity of AI-assisted eye disease diagnosis technology should be more than 90%. In addition, deep learning model under doctors’ supervision is the best choice in the practice of community-based eye disease screening.