AI Case Study
SocialEyes diagnoses diseases in places where doctors are scarce by scanning the human retina with deep neural networks at the edge
The human retina contains a wealth of diagnostic markers for many diseases such as diabetes and hypertension. There is a significant shortage of eye doctors globally and most are based in urban settings meaning that many diseases go undiagnosed where doctors are scarce. SocialEye's mobile, autonomous, retinal evaluation (MARVIN) system uses 'edge' neural networks to diagnose diseases in real-time. This allows health workers in rural settings to better manage diseases that all create problems with vision.
Healthcare Equipment And Supplies
"SocialEyes MARVIN (for mobile autonomous retinal evaluation) –which runs on GPU-powered Android tablets – helps community health workers manage eye problems caused by diabetes, hypertension and cardiovascular disease close to home. Untreated, these lead to a host of debilitating problems, including blindness.
SocialEyes’ GPU-enabled image processing and machine learning software spots 'signatures' of retinal problems caused by diabetes and related diseases.
An app can assess the retina’s condition and treatment can begin immediately, rather than waiting days for a report to come back from a remote tele-health grading center. By working offline—without an Internet connection—MARVIN allows local management of symptoms that would otherwise trigger an automatic and costly specialist referral."
"The human retina contains a wealth of diagnostic markers for many non-communicable diseases, such as diabetes and hypertension, as well as infectious conditions such as tuberculosis, HIV, dengue and malaria.
Eye care in developing countries occurs in overflowing general hospitals and clinics, as well as 'outreach camps,' where hundreds to thousands of patients are seen over a few days.
There are 200,000 eye doctors worldwide who are located primarily in urban setting. And there are one billion persons with diabetes, hypertension and other diseases all face problems with their vision."
"Most new cases are occurring in low-income countries, where resources are strained and many people aren’t even aware of their condition until a crisis occurs."
"With MARVIN, doctors, ophthalmic assistants and community health workers can now make informed decisions in moments, while the patient is there. This encourages personalized counseling and education, as well as integrating advanced eye care with treating the systemic disease."
"By working offline—without an Internet connection—MARVIN allows local management of symptoms that would otherwise trigger an automatic and costly specialist referral."
"Although the world’s supply of eye specialists is limited, there are over 15,000,000 general practitioners—and even more nurses, support staff and community health workers—who can use SocialEyes and MARVIN to manage their patients in primary care."
"SocialEyes technology, running on mobile supercomputer tablets, makes key retinal features easier to see. Neural nets and machine learning help ensure that high-risk cases are caught right at the point of initial contact."
"Marginal image quality is common in low-resource settings, especially with non-mydriatic (undilated) imaging."
It is assumed that the deep neural networks would have been trained on images of retinal scans labelled with conditions.