Artificial intelligence using deep learning to screen for referable and vision-threatening diabetic retinopathy in Africa: a clinical validation study.

BACKGROUND: Radical measures are required to identify and reduce blindness due to diabetes to achieve the Sustainable Development Goals by 2030. Therefore, we evaluated the accuracy of an artificial intelligence (AI) model using deep learning in a population-based diabetic retinopathy screening programme in Zambia, a lower-middle-income country. METHODS: We adopted an ensemble AI model consisting of a combination of two convolutional neural networks (an adapted VGGNet architecture and a residual neural network architecture) for classifying retinal colour fundus images. We trained our model on 76 370 retinal fundus images from 13 099 patients with diabetes who had participated in the Singapore Integrated Diabetic Retinopathy Program, between 2010 and 2013, which has been published previously. In this clinical validation study, we included all patients with a diagnosis of diabetes that attended a mobile screening unit in five urban centres in the Copperbelt province of Zambia from Feb 1 to June 31, 2012. In our model, referable diabetic retinopathy was defined as moderate non-proliferative diabetic retinopathy or worse, diabetic macular oedema, and ungradable images. Vision-threatening diabetic retinopathy comprised severe non-proliferative and proliferative diabetic retinopathy. We calculated the area under the curve (AUC), sensitivity, and specificity for referable diabetic retinopathy, and sensitivities of vision-threatening diabetic retinopathy and diabetic macular oedema compared with the grading by retinal specialists. We did a multivariate analysis for systemic risk factors and referable diabetic retinopathy between AI and human graders. FINDINGS: A total of 4504 retinal fundus images from 3093 eyes of 1574 Zambians with diabetes were prospectively recruited. Referable diabetic retinopathy was found in 697 (22·5%) eyes, vision-threatening diabetic retinopathy in 171 (5·5%) eyes, and diabetic macular oedema in 249 (8·1%) eyes. The AUC of the AI system for referable diabetic retinopathy was 0·973 (95% CI 0·969-0·978), with corresponding sensitivity of 92·25% (90·10-94·12) and specificity of 89·04% (87·85-90·28). Vision-threatening diabetic retinopathy sensitivity was 99·42% (99·15-99·68) and diabetic macular oedema sensitivity was 97·19% (96·61-97·77). The AI model and human graders showed similar outcomes in referable diabetic retinopathy prevalence detection and systemic risk factors associations. Both the AI model and human graders identified longer duration of diabetes, higher level of glycated haemoglobin, and increased systolic blood pressure as risk factors associated with referable diabetic retinopathy. INTERPRETATION: An AI system shows clinically acceptable performance in detecting referable diabetic retinopathy, vision-threatening diabetic retinopathy, and diabetic macular oedema in population-based diabetic retinopathy screening. This shows the potential application and adoption of such AI technology in an under-resourced African population to reduce the incidence of preventable blindness, even when the model is trained in a different population. FUNDING: National Medical Research Council Health Service Research Grant, Large Collaborative Grant, Ministry of Health, Singapore; the SingHealth Foundation; and the Tanoto Foundation.

Prevalence of diabetic retinopathy and visual impairment in patients with diabetes mellitus in Zambia through the implementation of a mobile diabetic retinopathy screening project in the Copperbelt province: a cross-sectional study.

AIMS: A paucity of literature exists on prevalence of diabetic retinopathy (DR) in sub-Saharan Africa. We aim to estimate the prevalence of DR and visual impairment in Zambia's Copperbelt province through a cross-sectional study. METHODS: All patients with a diagnosis of diabetes mellitus attending a DR screening programme were eligible to participate. Fundus photographs were graded in accordance with the DR grading system used in the UK National Health service (NHS). Visual impairment data were collected from visual acuity measurements recorded using Snellen chart. RESULTS: A total of 2689 patients were screened and of these, 2153 patients had a least one eye of gradable quality for analysis. Fifty-five per cent (1190/2153) of patients were male. Mean age was 56 (SD 11). Fifty-two per cent (1113/2153) showed evidence of diabetic retinopathy (DR). Thirty-six per cent of patients graded (779/2153) had sight threatening DR. Proliferative DR was found in 7% (14/208) of type 1 diabetics compared to 5% (42/921) type 2 diabetics (p = <0.001). Duration of diabetes, random blood glucose, systolic and diastolic BP, and use of insulin and oral hypoglycaemics were strongly associated with DR in univariate analysis. The associations of increased systolic BP, random blood glucose, duration of diabetes and insulin use with DR were maintained in multivariate analysis. CONCLUSION: We observed a high prevalence of sight threatening DR which is close to the upper range of estimates that currently exist on DR. This study represents further evidence of global health inequality and the scale of the epidemic which sub-Saharan African countries now face.

Outcomes of Pediatric Cataract Surgery in the Copperbelt Province of Zambia.

PURPOSE: To study the outcomes of pediatric cataract surgeries in children operated on in the Copperbelt Province of Zambia and the barriers to accessing surgery. METHODS: All children who had congenital, developmental, and traumatic cataracts operated on by lens aspiration, primary posterior capsulotomy, and anterior vitrectomy with posterior chamber intraocular lens implantation from 2012 to 2013 and followed up beyond 6 months were studied. Each child underwent a comprehensive preoperative evaluation. An active, assisted follow-up was done and parents were asked about reasons for delay between presentation and surgery. RESULTS: One hundred two eyes of 70 children met the inclusion criteria of the study. Preoperatively, 76 of 77 (98.7%) eyes in the congenital and developmental cataract group had presenting visual acuity of worse than 6/60. This improved postoperatively, with 19 (29.7%) eyes having best corrected visual acuity (BCVA) of better than 6/18, 23 (35.9%) having BCVA of 6/24 to 6/60, and 22 (34.4%) having BCVA of worse than 6/60. Older age (P = .005), better preoperative vision (P = .045) at presentation, unilaterality (P = .012), and delay between presentation and surgery (P = .004) were predictors of a better postoperative outcome. On multivariate analysis, only age was significant (P = .025). Distance and cost of travel and surgery were the causes of delay in presentation. In the patients with traumatic cataracts, 17 of 25 (68%) had BCVA of better than 6/18, 6 (24%) had BCVA between 6/24 and 6/60, and 2 (8%) had BCVA of worse than 6/60 at the 6-month follow-up. The most common causes of injury were being struck by sticks and stones (10 children, 25%). CONCLUSIONS: Visual outcomes after pediatric cataract surgery in Zambia were comparable and satisfactory. Cost of treatment was a barrier, but delay did not adversely affect outcome. [J Pediatr Ophthalmol Strabismus. 2016;53(5):311-317.].