Health and Wellness
With reference to the use of artificial intelligence (AI) diabetes scanners in the province:
- (a)(i) How many facilities currently have access to AI-based diabetes scanning technology and (ii) where are these facilities located, (b) what outcomes have been recorded from the use of AI-based diabetes scanners in the province and (c) how are these outcomes measured;
- whether any success stories have been documented regarding the use of AI-based diabetes scanners; if so, what are the key findings;
- (a) what impact has the use of AI-based diabetes scanners had on early detection and management of diabetes and (b) how does this compare with traditional screening methods;
- (a) what plans are in place to expand access to AI-based diabetes scanning tech-nology and (b) what targets have been set in this regard?
(1)(a)(i) Four primary healthcare facilities currently have access to AI-assisted diabetic retinopathy screening technology.
(1)(a)(ii) These facilities are located as follows:
Khayelitsha subdistrict: Michael Mapongwana Community Health Centre
Khayelitsha subdistrict: Khayelitsha Site B Clinic
Khayelitsha subdistrict: Nolungile Clinic
Eastern subdistrict: Mfuleni Community Day Centre
The service is currently implemented in the Khayelitsha/Eastern Substructure in partnership with Orbis International.
(1)(b) The key outcomes recorded include improved access to diabetic eye screening at primary healthcare level, increased screening coverage, improved detection of diabetic retinopathy, faster grading of retinal images, and more timely referral of patients requiring further assessment or treatment.
Screening increased from 1 125 patients before the introduction of AI-assisted screening to 5 012 patients in year 1 and 8 376 patients in year 2. The number of detected diabetic retinopathy cases increased from 189 before AI-assisted screening to 1 598 in year 1 and 2 512 in year 2.
(1)(c) Outcomes are measured through routine monthly facility reporting, AI-generated grading outputs from retinal images, programme monitoring and evaluation, and indicators such as the number of patients screened, the number of patients diagnosed with diabetic retinopathy, referral rates, and treatment uptake.
(2) Yes. Success has been documented. The programme has demonstrated that AI-assisted diabetic retinopathy screening can increase screening coverage, support earlier detection, and bring screening closer to communities at primary healthcare level. The programme has also been recognised as a service of excellence and has supported broader learning around the use of AI in public sector healthcare.
An anonymised patient example also demonstrates the value of the service. A patient who did not have noticeable symptoms was screened through the AI-supported service and found to have advanced diabetic retinopathy. This enabled referral for further care and helped reduce the risk of further vision loss.
(3)(a) The use of AI-assisted screening has strengthened early detection by enabling retinal images to be graded rapidly at the point of care. This supports earlier identification of diabetic retinopathy, quicker referral where required, and better integration of eye screening into routine diabetes care.
(3)(b) Compared with traditional screening methods, AI-assisted screening helps address delays linked to limited specialist capacity and the need for manual grading. It also enables trained non-specialist staff to conduct screening closer to where patients receive routine diabetes care, while still supporting appropriate referral pathways for patients who require further assessment or treatment.
(4)(a) The Western Cape Department of Health and Wellness supports the further expansion of AI-assisted diabetic retinopathy screening, subject to available resources and procurement processes. Expansion would require the procurement of additional equipment to extend the service to more facilities within the Khayelitsha Eastern Substructure and potentially to other areas.
(4)(b) No final quantified expansion targets have been set at this stage. Any future targets will be informed by available funding, procurement of additional equipment, service readiness, and lessons from the current implementation sites. The broader aim is to increase the number of facilities offering AI-assisted screening, increase the number of patients screened annually, improve early detection, and reduce avoidable vision loss among people living with diabetes.
