Artificial Intelligence-Supported Diabetic Retinopathy Screening in Tanzania: A Randomised Controlled Trial

The early diagnosis and treatment of diabetic retinopathy (DR) is known to preserve vision; this is done successfully in the UK through the national DR screening programme. However, there are few DR screening programmes in Africa, which have little quality control and high rates of loss to follow-up.

This project will randomise persons eligible for DR screening to either the standard service model or an artificial intelligence (AI) supported screening model which provides an immediate diagnosis and therefore allows point of screening patient counselling.

We will assess whether AI-supported DR screening can lead to improved patient outcomes within two real-world screening programmes in Tanzania.

Project Aims

The aim of this research project is to assess whether AI-supported diabetic retinopathy screening increases the proportion of persons with true referable diabetic retinopathy who attend the central ophthalmology clinic through a “double-boost” effect by

  • improving grading accuracy

AND

  • facilitating patient counselling at the point of screening following an immediate referral decision.

This will test whether the implementation of an AI-supported DR screening system leads to better quality care for persons screened for DR and improved outcomes through higher rates of ophthalmology follow-up for those with referable retinopathy.

Specific Aims:

  1. Conduct a scoping review and situational analysis of DR screening programmes in SSA and the current evidence on the use of AI for DR screening and management in LMICs.
  2. To determine whether AI-supported DR screening increases the proportion of persons with true referable retinopathy who attend ophthalmology follow-up.
  3. To compare prospective programmatic sensitivity and specificity of AI-supported and standard fundus image grading.
  4. To determine the acceptability, sustainability and fidelity of AI-supported DR screening.
  5. To perform a cost comparison between AI-supported and standard DR screening.

Study Sites and Participants

This study will be conducted in 2 centres in Tanzania. The study sites are the Kilimanjaro Christian Medical Centre (KCMC) in Moshi and Muhimbili Hospital in Dar es Salaam.

The participants in the study will include persons with diabetes enrolled in the DR screening programmes based at the 2 study sites. We estimate enrolling about 2000 participants over 18 months.

Associated Publications

  1. Cleland CR, Burton MJ, Hall C, Hall A, Courtright P, Makupa WU, et al. Diabetic retinopathy in Tanzania: Prevalence and risk factors at entry into a regional screening programme. Trop Med Int Heal. 2016.
  2. Cleland CR, Burton MJ, Hall C, Hall A, Courtright P, Makupa WU, et al. Diabetic retinopathy screening: Experiences from northern Tanzania. The Lancet Diabetes and Endocrinology. 2016.
  3. Mtuya C, Cleland CR, Philippin H, Paulo K, Njau B, Makupa WU, et al. Reasons for poor follow-up of diabetic retinopathy patients after screening in Tanzania: A cross-sectional study. BMC Ophthalmol. 2016.

Acknowledgments and Funding

Funded by British Council for the Prevention of Blindness (BCPB).

Dr Bernadetha Shilio at the Ministry of Health Tanzania.

Ophthalmology staff at KCMC and Muhimbili Hospitals, Tanzania.

Specific Contacts

Charles Cleland charles.cleland@lshm.ac.uk

Prof. Matthew Burton matthew.burton@lshtm.ac.uk

Cova Bascaran Covadonga.bascaran@lshtm.ac.uk