Researchers at Moorfields Eye Hospital and UCL Institute of Ophthalmology (IoO) have developed an artificial intelligence (AI) model called RETFound that can predict heart attack risks by examining a patient's eyes.
The UCL Institute of Ophthalmology is part of the Faculty of Brain Sciences of University College London.
The RETFound technology looks at a person's retina to predict their heart attack risk factors. It leverages vast amounts of National Health Service image and text data to screen patients for other diseases, including diabetic eye disease, glaucoma, Parkinson's disease and stroke. This technology can identify serious medical conditions in many patients before they exhibit symptoms, leading to early intervention and treatment and the potential to save many lives.
RETFound and other similar technologies are part of the broader oculomics field, an emerging area of study that examines the eye to spot biomarkers of various diseases and conditions.
Pearse Keane, a professor of artificial medical intelligence at UCL, said, "I think a big shortcoming of current AI systems is, if you have a model developed in London and maybe it's trained on photos from a white Irish guy in his 40s, then maybe that's not going to work as well on a patient in Ghana with diabetic eye disease. This is about trying to make these AI systems robust and reliable and safe and fair for the future."
An ophthalmologist in Tanzania who wants to develop an AI system to perform diabetic retinopathy screening of the local population could get UCL's system and fine-tune it with a small amount of East African data to create a model that potentially works in that setting, Keane said.
RETFound works by analyzing images of the retina to look for subtle changes that may indicate the presence of underlying disease. The model trains on a large dataset of images from patients with and without a variety of diseases. It learns to identify patterns associated with different diseases, improving over time because of the quantity and quality of the data.
Once the model receives a large number of training scans, it can screen new patients for disease by receiving photographs of a patient's retina into the AI model. The AI analyzes the images and compares them to others within its database and then outputs a probability score that indicates the patient's risk of having each of the potential trackable diseases.
RETFound can revolutionize disease detection and treatment by identifying people who are at risk of developing serious diseases before they experience symptoms. It can also improve the efficiency of healthcare systems by, for example, screening patients during their routine eye exams. This could reduce the number of patients who need specialist referrals for further investigation and ensure early treatment interventions happen for other patients.
Despite the promise of the RETFound solution, challenges remain. The model needs validation on a large, independent dataset to ensure that it is accurate and reliable. Another challenge is that the model needs to integrate within existing National Health Service (NHS) systems so clinicians can leverage AI in their everyday practices.
Starpax Biopharma is a startup that is also using the latest technology to save lives. It offers a cancer treatment platform that uses "magnetodrones" to deliver Food and Drug Administration (FDA)-approved anticancer drugs directly into tumors for maximum results.
