Scientists create AI tool that anticipates disease – 07/21/2023 – Health

The upright posture of humans is an evolutionary step that still intrigues many scholars of great apes. The fact that we walk bipedally all the time challenges human bone development, especially in adulthood when the risk of fractures can lead to bone damage and other conditions.

Genetic factors may also play a role in the onset of these conditions, but this was poorly understood until then. Now, new research has revealed that genetic predisposition, combined with anatomical factors such as bone size ratio, can anticipate future bone damage.

The study, published on the cover of Science this Thursday (20), brings new data about our evolutionary past and can help doctors to better predict the risk of patients developing conditions in adult life, such as back pain, knee pain and arthritis.

Using artificial intelligence to analyze thousands of X-ray images and information from our DNA, researchers at the University of Texas at Austin and the New York Genomic Center have identified the genes involved in the development of our skeletons, from shoulder width to leg length.

“We found that high skeletal proportions related to the knee joint [uma maior razão entre comprimento e altura do fêmur, da tíbia e do ângulo tibiofemoral] were associated with a higher incidence of arthritis and knee pain. Furthermore, a greater ratio of hip width to height was linked to an increased risk of osteoarthritis and hip pain,” said Eucharist Kun, a researcher in the Department of Integrative Biology at the University of Texas and first author of the study.

Likewise, people with a higher ratio of torso height to hip width are at greater risk of developing back pain. This means, in general, that people with these characteristics can suffer from these diseases in the future, but genetic predisposition is not the only factor.

Humans are the only great apes to have legs longer than their arms, a fundamental skeletal change to ensure bipedal walking. But analysis of the fossil record alone, from non-bipedal hominids to modern humans, has not been sufficient to understand the genetic changes behind these different anatomical shapes.

In the study, scientists analyzed 39,172 full-body X-ray images of individuals aged 40 to 80 years from the UK Biobank, one of the largest biomedical repositories. Using a “deep learning” tool allowed the scientists to measure 14 distances between body parts, such as the distance between the shoulders, knees and ankles.

The real discovery happened when the scientists correlated these measurements with the genetic sequences of each individual, identifying 145 points in the genome (alleles, or the copies that each person carries of a determined gene) responsible for controlling the skeletal proportions.

“Our research is a powerful demonstration of the impact of artificial intelligence in medicine, especially when it comes to analyzing and quantifying image data, as well as integrating that information quickly and at scale with health records and genetics,” said Vagheesh Narasimhan, assistant professor of integrative biology and statistics at the University of Texas, who led the study.

Some of the measures associated with the genes may be an indicator of the onset of musculoskeletal diseases in the future, such as arthritis of the knee and hip —conditions that affect billions of people in the world and are highly disabling in adult life.

“Although we cannot say with certainty that a certain region of the DNA corresponds to an increased risk of skeletal diseases, what we can say is that the relationship between specific measures of the skeleton [como relação do tamanho do torso com o quadril] may be associated with the development of pain”, evaluates Kun.

This information can be useful for the diagnosis and medical examination of patients, completes the scientist. “Routine examination of bone conditions involves the use of X-rays. With our pioneering method of analyzing distances between bones using AI, we believe this could be a potential tool to aid in clinical diagnosis.”

In addition to medical use, the results of the work have implications for evolutionary studies. That’s because scientists have identified areas in the genome known as “human accelerated genomes”, areas of DNA conserved throughout vertebrate evolution, including within the primate group, but which have suffered a significant number of mutations in humans.

“Our study revealed that the regions of the genome and the genes associated with skeletal proportions are favored together with these human accelerated regions, especially those linked to bipedal walking”, says the researcher.

Since Leonardo da Vinci’s studies on the proportions of the limbs and other elements of the human body, represented in the famous figure of the “Vitruvian Man”, scientists have sought to understand the evolutionary factors involved in human anatomy.

“In a way, we’re tackling the same question da Vinci faced: what is the basic human form and its proportions? But now we have modern methods and also asking the AI ​​tool how these proportions are determined genetically.”