Research updates & events
Updated: 1 hour ago
Over the past few months, members of Pearse Keane’s AI lab at UCL Institute of Ophthalmology, and our wider network of clinical research colleagues across Moorfields Eye Hospital and other institutions, have published a number of noteworthy studies on varied facets of health data analysis, deep learning, AI development and implementation. Read on for a selection of papers, other news, and key events where the team will be presenting.
Notable papers
Oculomics
People with atrial fibrillation (AF) have distinctive biomarkers in their retinas, evident as early as four years before clinical presentation of the condition, confirms a new study published in PLOS Digital Health, led by Josef Huemer and Siegfried Wagner. The research has received widespread coverage in the UK media. Read more
The UK is uniquely placed to realise the potential of Oculomics, and translate eye biomarkers into population health, argues a new commentary in Eye by Pearse Keane, Siegfried Wagner and Alexander Heatley.
New research published in Nature Portfolio Mental Health outlines a roadmap for Oculomics in psychiatric research, the study of eye-based biomarkers for assessing mental health conditions.
Evidence is growing for the use of AI-based retinal imaging to assess brain health. As featured on the cover of The Lancet Digital Health, our colleagues undertake a comprehensive review of state-of-the-art developments, identify gaps, and propose directions for future research to facilitate the translation of these concepts into real-world applications.
Ophthalmology
As part of ongoing collaborative work between the Keane lab and Massachusetts Eye and Ear, Ariel Ong and Ines Lains led research examining the association between reticular pseudodrusen and dark adaptation through a novel quantitative approach using an OCT-based deep learning algorithm. The prospective cross-sectional and longitudinal study has been published in Ophthalmology Retina.
A study in Ophthalmology Science updates decades-old research looking at vascular morphology with incident DR progression, but scales the data size to 20 times that of prior studies, with the Automorph tool. It also uses more clinically relevant imaging to give more precise estimates for the associations.
Implementation
A new Nature Portfolio, authored by Yilan Wu and Pearse Keane together with Alvin Liu at Johns Hopkins Medicine and Eric Topol at Scripps Research, argues that the integration of health AI systems into the care pathway is creating a significant risk that warrants greater scrutiny.
An analysis of how AI and telemedicine are reshaping ophthalmology patient journeys in the United Kingdom and Germany shows that the same AI technologies can lead to very different care pathways across health care systems. The paper has been published in the Journal of Medical Internet Research.
AI development
Most ophthalmic AI foundation models are developed using data from North America, Western Europe, and Southeast Asia, with an estimated 3.5 billion people worldwide underrepresented in open-access AI datasets. A new commentary and call-to-action in The Lancet Regional Health – Africa argues for concentrated efforts to close this AI equity gap.
Federated learning in healthcare allows researchers from different organisations to build a joint AI model across borders. However, a new commentary in Nature Portfolio highlights that only 5.2% of published healthcare federated learning studies result in real-world clinical application, with the vast majority being proofs of concept or simulated experiments run on data already centralized in one place.
Large language models
In their paper, “Fast information and slow evidence in the large language models era” Keane lab alumni highlight how LLMs can support evidence infrastructures, clarify evidence boundaries, and reshape clinical expertise to strengthen rather than replace evidence-based medicine. Published in npj medicine.
Writing for The Lancet Digital Health, Ariel Ong and co-authors discuss the theoretical underpinnings of “co-intelligence” and its role in medical research. They highlight key considerations, discuss potential unintended consequences, and provide insights for researchers and clinicians who want to maximise the use of LLMs while maintaining the rigour and integrity of their work.
In other updates
For researchers working on human-centred artificial intelligence, npj Digital Medicine is seeking submissions for a special collection edition, co-curated by INSIGHT Data Lead Ariel Ong. Submission deadline: 3rd September 2027 Find out more.
An AI tool for predicting risk of diabetic retinopathy has progressed to a 2-year 'silent trial' in the NHS - the first of its kind in the UK. The trial builds on AI development work led by Paul Nderitu, Tim Jackson and Christos Bergeles, validated with data from INSIGHT Eye Hub and supported by Sight Research UK. Read about the trial in Eye News.
Congratulations to INSIGHT emeritus director Alastair Denniston on being appointed to lead the NIHR (National Institute for Health and Care Research) Health and Social Care Delivery Research programme. Read more on the NIHR website.
Medical AI spinout Cascader Ltd has secured seed funding from Topcon Healthcare to progress development of products in eye health and Oculomics. Cascader Ltd was established through a partnership between Moorfields Eye Hospital, UCL Ventures and Topcon Healthcare. INSIGHT is supporting AI development at Cascader. Read more on the Cascader website.
Pearse Keane was a recent guest on the Age of AI podcast, hosted by Lara Lewington and Ben Clark. Watch on YouTube.
Where to find us
DOG Kongress (German Society of Ophthalmology), Berlin, 24th-27th September 2026
Siegfried Wagner is presenting on “Early detection of neurologic and neuropsychiatric diseases based on AI-supported fundus imaging”
EURETINA, Vienna, 1st - 4th October 2026
Pearse Keane, Siegfried Wagner, Ariel Ong and Paul Nderitu are speaking across multiple sessions. Topics include: Mapping eye-brain relationships with cross-modal AI;
From Retina to Brain: How AI is Transforming Our Understanding of Neurodegeneration;
Foundation Models: Opportunities for Surgical Intelligence; How can we use -omics in neonatal care?; Oculomics, neurodegeneration and AI.
American Academy of Ophthalmology annual meeting, New Orleans, 9th-12th October
Pearse Keane is speaking on scaled sustainable care in AI and tele-ophthalmology, and on taking AI from Research to real-world workflow in ophthalmology
Future of Eye Care conference, London, 17th October 2026, hosted by Moorfields Eye Charity and Friends of Moorfields
In the morning session, Siegfried Wagner, Roxanne Crosby-Nwaobi and Paul Nderitu are presenting on impactful ways to reduce health data poverty, culminating in a panel discussion with Moorfields colleagues. INSIGHT will also have a stand at the event.
Asia-Pacific Ophthalmic Innovation Society (APOIS) conference, Seoul, 30th October - 1st November 2026
Paul Nderitu and Yukun Zhou are presenting pivotal analysis results of the Global RETFound project. Siegfried Wagner will also be speaking.
ISPOR Europe, Vienna, 8th - 11th November 2026
Ariel Ong is speaking on the use of LLMs for data extraction to support real-world evidence generation, and on patient engagement in economic modelling
FLORetina, Florence, 3rd-6th December 2026
Pearse Keane is leading a course on “AI for Oculomics Research: Retinal Imaging as a Window to Systemic Health”
RCOphth Winter Meeting, London, 9th December 2026
Ariel Ong, Josef Huemer and Tom Taylor will present on various aspects of AI in ophthalmology
