diabetic retinopathy diabetic macular edema artificial intelligence ARDA diabetic retinopathy diabetic macular edema screening
Conditions
Interventions
Artificial intelligence as a DR screening method. According to previous study by Gulshan et al,
ARDA,
the artificial intelligence,
has 99.0%-
99.1% accuracy,
87.0%-
90.3%sensitivity and 98.1-
98.5% specificity.
Artificial intelligence as a DR screener
Sponsors
Rajavithi hospital
Google Inc.
Eligibility
Sex/Gender
All
Age
0 Years to 0 Years
Inclusion criteria
Inclusion criteria: - Patients with diabetes and had their retinal images taken in the National DR Screening Program in Ophthalmic Service Plan of Ministry of Public Health of Thailand
Exclusion criteria
Exclusion criteria: - Patients with other retinal diseases (age-related macular degeneration, retinal vein occlusion, and other retinal vascular diseases) will be excluded by retinal specialists - Patients with inability of retinal images taken by digital fundus camera - Patients who are unwilling to attend the study
Design outcomes
Primary
| Measure | Time frame |
|---|---|
| To evaluate efficacy of Deep learning system in screening diabetic patients in communities to detect after complete 7600 patients AI screening comparing to retinal specialists | — |
Secondary
| Measure | Time frame |
|---|---|
| To find 1-year incidence of sight-threatening diabetic retinopathy by conventional method and deep l 1 year AI screening comparing to retinal specialists,To find 1-year prevalence of sight-threatening diabetic retinopathy by conventional method and deep after complete 7600 patients retinal specialists,To find compliance of diabetic retinopathy screened by deep learning system in visiting ophthalmolog after complete 7600 patients follow up referal status,To assess mean time from screening to visiting ophthalmologist. after complete 7600 patients follow up referal status,To assess failure rate of deep learning system (delay fundus photo interpretation by deep leaning sy after complete 7600 patients follow up referal status,To assess accuracy of deep learning system in diabetic retinopathy screening program. after complete 7600 patients follow up referal status,To assess number of patients screened by deep learning system. after complete 7600 patients follow up referal status,To assess health care provider’s and patient’s satisfaction of deep learning system after complete 7600 patients follow up referal status | — |
Countries
Thailand
Contacts
Public ContactPaisan Ruamviboonsuk
Retinal specialist
Outcome results
None listed