The Nuffield Department of Primary Health Care Sciences is delighted to celebrate the achievements of James Leigh, a graduate of the MSc in Applied Digital Health and now a DPhil in Translational Health Sciences candidate within the department. James has successfully published his MSc dissertation as a peer-reviewed journal article in Clinical & Experimental Ophthalmology - a significant milestone for any early-career researcher and a testament to the impact of his work.
James’ research focuses on the cost-effectiveness of artificial intelligence (AI) systems used to screen for diabetic retinopathy (DR), a leading cause of blindness that affects approximately one in three people living with diabetes worldwide. While AI-enabled screening technologies have shown accuracy comparable to that of trained clinicians, questions remain around their real-world value, scalability and economic viability across diverse healthcare settings. His review addresses these questions at a critical moment, as health systems worldwide increasingly explore the use of AI to support earlier detection and improved patient outcomes.
In his paper, James examined 18 economic evaluations comparing AI-assisted diabetic retinopathy screening with traditional manual grading by clinicians. He and his co-authors identified three key considerations when determining if AI represents value for money. These findings have wide relevance for policymakers, clinicians and technology developers.
First, the review highlights the importance of local context for cost-effectiveness. Economic evaluations must account for regional variations in healthcare costs, infrastructure and patient pathways. For example, in rural or remote areas the higher costs associated with false-positive results, including long travel distances for patients and lost income, can significantly influence whether a particular AI system is cost-effective. The varying costs of false-positive and false-negative results among patients should be considered to determine the optimal balance between an algorithm’s sensitivity and specificity.
Second, the study underscores the critical role of real-world data on referral uptake after DR screening. Emerging evidence suggests that AI screening pathways, which can provide immediate results to patients, may improve follow-up rates compared to conventional models of care. Yet most existing economic evaluations assume equivalent follow-up rates regardless of screening method. James and his co-authors argue this parameter should be integrated into modelling studies of AI-enabled screening programmes.
Finally, the review identifies the need for stronger reporting standards among economic evaluations of AI-enabled DR screening. Only 39% of the included studies followed standardised reporting guidelines. As AI becomes increasingly integrated into healthcare, James recommends that future health economic evaluations should adopt AI-specific frameworks such as CHEERS-AI and PICOTS-ComTeC to improve transparency and comparability of findings.
His article has also been selected by Clinical & Experimental Ophthalmology as one of five key papers to be highlighted in an upcoming issue - a recognition that celebrates both the quality and relevance of his work.
James is continuing to advance this research in his current DPhil studies, with a focus on AI-enabled DR screening across metropolitan, rural and remote communities in Western Australia. His work aims to support the development of safe, equitable and cost-effective screening pathways that can reduce preventable blindness and strengthen health outcomes for people living with diabetes.
We congratulate James on this outstanding achievement and look forward to following the next steps of his research journey.
You can read James’ paper now published in Clinical & Experimental Ophthalmology here.