WHAT YOUR EYES CAN TELL YOU ABOUT THE HEART
The fact that the eyes aren’t just about seeing things – they are, for example, part of the mechanism behind the biological clock – is something on which I’ve already written. They can also be examined to diagnose various diseases, even those that don’t involve the eyes at all.
This will soon be algorithmized – but more precisely, it has already been done. Google and Verily developed an algorithm that can predict cardiovascular diseases through the eyes.
Using a retinal scan, the computer algorithm can calculate fairly accurate data about your age, blood pressure, or if you smoke. The findings are used to predict heart attack.
Interestingly, the algorithm makes it possible to detect a problem without blood analysis.
So how did they achieve this? In order to make a model, a computer program analyzed the medical parameters of over 300,000 patients. Then a model was created that linked all that data with a metric for predicting cardiovascular disease.
The fact is that there are a multitude of blood vessels in the retina that reflect not only the state of the eyes themselves, but the body as a whole. That’s why ophthalmologists can “guess” the presence of problems related to blood pressure, diabetes, and a number of other diseases, not to mention smoking addiction.
So how effective is this algorithm? Pretty impressive – 70%, especially when compared to the SCORE method of blood analysis, which is at 72%. SCORE, however, is much more time-consuming.
Will we live to a point where AI can diagnose without the help of a doctor? I think it all depends on longevity – we’re talking decades. But there’s no more thankless job out there than predicting the future. For example, a neural network developed by Cardiogram can already diagnose diabetes with 85% accuracy. And it does this all without medical devices – all it needs is an Apple Watch or similar smart device.
Diagnosis and treatment are interrelated, but not the same thing. Treatment in the near future will still not come from a machine. But in the search for a better doctor, perhaps some algorithms could help, if they would take into account real responses from patients.