Divination by the eyes

Retinal aging was associated with an increased risk of death from all causes

Anastasia Kuznetsova-Fantoni , N+1

Doctors from three countries found a correlation between an increase in the age of the retina and an increased risk of mortality from all causes. To determine the retinal age of 35,000 people, they used an algorithm based on deep learning, and then compared the resulting value with the real age. After constructing a regression model, the scientists found that each additional year of the difference between the age of the retina and the chronological age of a person was associated with an increase in the risk of death from all causes by two percent. The work was published in the British Journal of Ophthalmology (Zhu et al., Retinal age gap as a predictive biomarker for mortality risk).

The condition of the retina can tell a lot about health. This is due to the fact that a large number of vessels are located on the fundus, in which pathological changes occur synchronously with changes in the vessels of the entire body. Therefore, ophthalmoscopy is often tried to adapt for the diagnosis of other diseases, even unrelated to vision.

For example, recently doctors have learned to determine the risk of cardiovascular diseases by the branching of retinal vessels and the condition of the optic nerve disc. They were helped in this by an algorithm based on deep learning methods, which, in addition to the risk of cardiac diseases, could also tell about the human body mass index, blood pressure and glycated hemoglobin.

Doctors from Australia, Germany and China, led by Mingguang He from the Hospital of Guangdong Province, used deep learning algorithms to predict the risk of death from all causes based on the branching and width of retinal vessels. First, the algorithm was trained to find out the age of a person by the condition of his fundus, using 19,200 images of the retina of the right eye from 11052 healthy people aged 40 to 69 years from the UK Biobank study. The algorithm has learned to name the age of a person with an average error of 3.55 years.

The researchers then took another 35,913 people from the UK Biobank study, for whom data on ten-year mortality were available. With the help of an already trained algorithm, doctors estimated the age of the retina of their eyes, and then compared it with the chronological age of the study participants. For statistical analysis, Cox regression was used to assess how the difference between retinal age and real age is related to mortality from all causes.

It turned out that each additional year of the difference between retinal age and chronological age is associated with an increased risk of death from all causes (hazard ratio=1.02, 95% CI 1.00-1.03, p=0.020). Interestingly, scientists have not found any correlation with the risk of mortality from cardiovascular and oncological diseases.

The work does not explain the reasons for the association found. In addition, the authors point to a limitation in the study — ophthalmoscopy was performed for each person only at the beginning of the study, and it was impossible to trace how the picture changed over time.

Machine learning methods are often superior to doctors in the diagnosis of diseases. So, American researchers have developed a method for the automatic diagnosis of neonatal retinopathy, which showed 98 percent accuracy and in the experiment surpassed the correctness of the diagnosis of professional ophthalmologists.

Portal "Eternal youth" http://vechnayamolodost.ru