The artificial intelligence It has been shown to be very effective in medicine. As much as to detect factors that human beings are incapable of discerning. For example, it has been predict the sex of a personsimply analyzing the retina in a photo of the fundus of your eye. It might seem like a banality were it not for the fact that, to this day, it is not known what are the characteristics that differentiate the retina of men and women. This does not have a great clinical implication, but it is a very clear example of what the deep learning can get.
To achieve this, an algorithm was trained. deep learning (deep learning) with 84,743 photos of the eye fundus from the UK Biobank. In addition, an external validation was performed with another 252 images from an ophthalmology reference center.
With both internal and external validation, the results were very good. The machine managed to capture what are those unknown factors that help to differentiate the eyes of men and women. There was just some trouble getting the AI to tell the sexes apart if the eye image belonged to someone with foveal pathology. However, even with this small limitation, it is shown that the deep learning it can reach even further than the complex rational mind of human beings. This has multiple applications and implications; but, focusing on eye healthIt is certainly a very positive thing.
Male or female eye
Physiologically there are many differences between people with male or female biological sex. Some are inconsequential, while others must be taken into account both for the disease diagnosis as for the decision of which are the best treatments. For example, when choosing a drug dosage, it is important to keep in mind that men and women may not tolerate the same amount.
In the case of the eyes there are also differences. In fact, the sex hormones they play a fundamental role in mechanisms such as eye hydration and circulation. Furthermore, the development of diseases What glaucoma, cataracts, or uveitis it can also be influenced by biological sex.
All these differences are becoming clearer. However, they are not known purely anatomical differences, that allow distinguishing if a photo of the eye fundus belongs to a man or a woman. At least, they are not known to human beings. For an artificial intelligence it may be a simpler matter. Just train an algorithm deep learning so that, based on collect images, it can find those differential points. And that is something that has already been achieved.
‘Deep learning’ to improve the diagnosis of ophthalmological diseases
The retina is the only tissue in which they can be visualized simultaneously and non-invasive both neural and vascular tissue.
This allows detecting a multitude of pathologies of both types, only with the analysis of images of the fundus of the eye. For example, vascular tortuosity and arterial narrowing allow detection of possible cardiovascular diseases. Instead, alterations in the retinal cell layer help diagnose Neurological disorders.
For all this, many studies have been carried out aimed at training artificial intelligence algorithms capable of exhaustively analyzing photos of the fundus of the eye. One of the most comprehensive studies on the subject It was carried out in 2018. Then, a team of scientists from Google and the Stanford School of Medicine managed to develop an algorithm that used the deep learning for predict cardiovascular risk factors in ocular fundus photographs. In his day, not only his results regarding the prediction of diseases drew attention. It also surprised the scientific community that the algorithm was able to distinguish between men and women. And that is something that scientists do not know how to do.
For this reason, another team of researchers, whose results were published in Nature in 2021, carried out that new training of deep learning with over 80,000 images from the UK Biobank. They managed to replicate the same as in 2018, so it did not seem like a simple coincidence. They were facing a new application of artificial intelligence.
But not precisely because it is useful to differentiate the retina of men from that of women. After all, when doctors receive a photo they have all the patient history, including information about your gender. What this study shows goes beyond all that. And that is how it shows that the deep learning has a great future ahead of it, in which new biomarkers associated with eye diseases could be brought to light. This does not mean that ophthalmologists will no longer be necessary. In fact, it is they, with their knowledge, who must guide the artificial intelligence on what to look for. Like other times, the perfect team is the one obtained with the union of the human and the machine.