Korean researchers have developed an algorithm using artificial intelligence to diagnose illness in children with 100% success. Its severity is also measured.
If artificial intelligence is often seen as a technological advance aiming to make our work easier (or to take it from us depending on the case), it is also used in many areas of research. There medicine is one of them. We have already seen ChatGPT find an illness that went unnoticed by 17 health professionals for example. But beyond this type of particular case, it is mainly at the level of diagnostics that AI proves its effectiveness.
In South Korea, researchers have published the findings of a study that could well mark a major breakthrough in early detection of disease in children. The team was interested in TSATHE autism spectrum disorders. Detecting them early allows better care, so she has developed an algorithm based onartificial intelligence to facilitate diagnostics. The results are very encouraging since the system is non-invasive and correct in 100% of cases.
Artificial intelligence diagnoses this disease in children with 100% success
As is often the case, it is the eyes that provide the necessary information. Or rather the retinas. Located at the back of the eye, it is connected to the optic nerve then the whole is attached to the optical disk. This structure, extension of central nervous systemis a real window into the brain. The researchers thus took the photo of the retinas of 958 children whose average age is 7.8 years. Of the 1,890 photos obtained, half belong to children who have already been diagnosed with ASD. The other half, without disorder, serves as a control group.
Also read – This AI looks you in the eyes and guesses your heart health
There severity of ASD was also measured using two dedicated scales. The AI was then trained with the already known photos and scores. After that, he was asked to detect ASD from retina photos And the AI made a correct diagnosis every time, without ever making a mistake. It is not as effective in determining the severity score, however, reaching a result considered acceptable, but not safe enough to rely on. The youngest participants were 4 years old. Before that, the retina is continuously growing, making earlier diagnosis difficult. The researchers are now working to obtain the generalizability of their results.
Source: JAMA Network Open