Colour fundus photography as a rapid screening tool for diabetic complications: a pilot study
1
Burden of Disease Research Unit, South African Medical Research Council, Cape Town, South Africa
2
Division of Epidemiology and Biostatistics, School of Public Health, University of Cape Town, Cape Town, South Africa
3
Somerset Eye, Somerset West, South Africa
Popul. Med. 2026;8(Supplement Supplement 1):A2932
ABSTRACT
BACKGROUND:
AI-based systems for interpreting colour fundus images are now regulator-approved in multiple countries for diabetic retinopathy screening and increasingly used in South Africa, where commercial solutions are reimbursed by major health insurers.[1] Evidence shows that retinal microvascular changes are biomarkers of systemic pathology. In diabetes, these alterations underlie complications such as cardiovascular disease, chronic kidney disease, and peripheral neuropathy, supporting the extension of AI-based retinal screening beyond retinopathy to detect other diabetes complications.[2,3]
METHODS:
We conducted a pilot study of 200 people with diabetes who underwent AI-based retinopathy screening in an endocrinology practice in the Western Cape between April 2023 and March 2025. Four colour fundus images, clinical histories, and relevant biomarkers were collected per participant. Retinal vascular features were quantitatively extracted and analysed using multivariate statistical methods to assess associations with systemic diabetic complications. A proof-of-concept deep-learning model was developed to predict early kidney disease from retinal images alone, with performance evaluated using standard classification metrics and attention-based analyses.
RESULTS:
Automatically extracted [4] measures of retinal vessel tortuosity, complexity, and density were associated with biomarkers of kidney function (eGFR, serum creatinine), glycaemic control (HbA1c), and cardiovascular risk (total cholesterol, LDL), as well as with clinical diagnoses of hypertension, kidney disease, and peripheral neuropathy. The screening model identified individuals at increased risk of kidney disease (eGFR < 60 mL/min/1.73 m² and/or albumin–creatinine ratio ≥ 3 mg/mmol), present in 19% of participants, with 87% sensitivity and an area under the curve (AUC) of 0.81.
CONCLUSIONS:
In people with diabetes, AI-derived retinal vascular features are associated with multiple systemic complications. These findings support the feasibility of extending retinal AI screening beyond retinopathy to enable non-invasive, multi-disease risk stratification. This approach could substantially expand screening for diabetic complications in South Africa, where current screening rates are well below clinical guideline recommendations.
AI-based systems for interpreting colour fundus images are now regulator-approved in multiple countries for diabetic retinopathy screening and increasingly used in South Africa, where commercial solutions are reimbursed by major health insurers.[1] Evidence shows that retinal microvascular changes are biomarkers of systemic pathology. In diabetes, these alterations underlie complications such as cardiovascular disease, chronic kidney disease, and peripheral neuropathy, supporting the extension of AI-based retinal screening beyond retinopathy to detect other diabetes complications.[2,3]
METHODS:
We conducted a pilot study of 200 people with diabetes who underwent AI-based retinopathy screening in an endocrinology practice in the Western Cape between April 2023 and March 2025. Four colour fundus images, clinical histories, and relevant biomarkers were collected per participant. Retinal vascular features were quantitatively extracted and analysed using multivariate statistical methods to assess associations with systemic diabetic complications. A proof-of-concept deep-learning model was developed to predict early kidney disease from retinal images alone, with performance evaluated using standard classification metrics and attention-based analyses.
RESULTS:
Automatically extracted [4] measures of retinal vessel tortuosity, complexity, and density were associated with biomarkers of kidney function (eGFR, serum creatinine), glycaemic control (HbA1c), and cardiovascular risk (total cholesterol, LDL), as well as with clinical diagnoses of hypertension, kidney disease, and peripheral neuropathy. The screening model identified individuals at increased risk of kidney disease (eGFR < 60 mL/min/1.73 m² and/or albumin–creatinine ratio ≥ 3 mg/mmol), present in 19% of participants, with 87% sensitivity and an area under the curve (AUC) of 0.81.
CONCLUSIONS:
In people with diabetes, AI-derived retinal vascular features are associated with multiple systemic complications. These findings support the feasibility of extending retinal AI screening beyond retinopathy to enable non-invasive, multi-disease risk stratification. This approach could substantially expand screening for diabetic complications in South Africa, where current screening rates are well below clinical guideline recommendations.
| eISSN: | 2654-1459 |
We process personal data collected when visiting the website. The function of obtaining information about users and their behavior is carried out by voluntarily entered information in forms and saving cookies in end devices. Data, including cookies, are used to provide services, improve the user experience and to analyze the traffic in accordance with the Privacy policy. Data are also collected and processed by Google Analytics tool (more).
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
You can change cookies settings in your browser. Restricted use of cookies in the browser configuration may affect some functionalities of the website.
