Asked to look at these four X-rays, the average person will see no difference. But AI can. In fact, research shows artificial intelligence is now so precise it can pinpoint the exact age of someone based solely on a chest scan.
The tool, developed by scientists in Japan, was fed images of thousands of actual radiographs and repeatedly tested and tuned.
Testing results showed the system could accurately predict age, with it being just three years out, on average.
Chest X-rays were gathered from 36,000 healthy patients scanned between 2008 and 2021. Scans from three sites were used to train and tune the AI bot. Results, published in The Lancet Healthy Longevity, showed that the tech’s guess of a person’s age correlated ‘very strongly’ with their actual age. Saliency graphs show that people aged 21 to 40 were far colder in the lower lung region than those aged 81 to 100. This is highlighted by the blurred blue patches. Those aged 81 to 100 show far more red areas around the heart which indicate it is much hotter. The chest radiographs also show how the chest cavity is typically much larger for those aged 21 to 40 when compared to people aged 81 to 100. The images appear blurred because they were created using downsampled images. Pictured: X ray scans (top) and saliency maps (bottom)
Experts also said that when it is wrong and overestimates, it is likely a sign that the patient has an illness.
Osaka Metropolitan University scientists say their findings marks a ‘leap in medical imaging’ and could pave the way for detecting and treating diseases such as high blood pressure, lung disease and kidney failure earlier.
Chest X-rays were gathered from 36,000 healthy patients scanned between 2008 and 2021.
Scans from three sites were used to train and tune the AI bot.
Results, published in The Lancet Healthy Longevity, showed that the tech’s guess of a person’s age correlated ‘very strongly’ with their actual age.
Researchers also showed it chest scans from another 34,000 patients who had a range of underlying conditions, such as asthma, diabetes and liver disease.
When the AI guessed someone was older than they actually were, it was a sign they were more likely to suffer from diseases like chronic obstructive pulmonary disease (COPD) and osteoporosis.
All of these conditions trigger changes in the chest, such as the blood vessels, lungs, bones and heart, that can be detected in an X-ray, the researchers said.
They looked at saliency maps — heat maps highlighting areas on the scan that the AI used to make its decision on the age of a patient.
These suggest that the bot was focusing on the mediastinum — the space that separates the lungs and holds the heart and vital arteries — to determine the age of a patient.
The scientists said looking at this area of the scan can reveal how much calcium has built up around the organs, with high levels being a marker of disease.
Yasuhito Mitsuyama, lead study author, said: ‘Our results suggest chest radiography-based apparent age may accurately reflect health conditions beyond chronological age.
‘We aim to further develop this research and apply it to estimate the severity of chronic diseases, to predict life expectancy, and to forecast possible surgical complications.’