Face Value
Patients may not have to undergo extensive clinical tests to diagnose some genetic disorders in the near future.
US company FDNA has developed facial recognition software that quickly identifies genetic syndromes by scanning a person’s face, the Verge reported.
In a recent study, researchers described how they trained the software – known as DeepGestalt – by feeding it 17,000 images of 200 different syndromes.
During tests, DeepGestalt’s algorithms accurately spotted specific disorders more than 90 percent of the time, such as the Cornelia de Lange syndrome and Angelman syndrome, both of which affect intellectual development and mobility. Medical experts got it right 70 percent of the time.
In another experiment, the software tried to identify the specific mutation causing a particular syndrome. This time, it was less accurate, with a hit rate of 64 percent.
Bruce Gelb, a medical professor who was not involved in the study, called the findings impressive but said traditional testing was still needed to identify specific mutations.
“It’s inconceivable to me that one wouldn’t send off the panel testing and figure out which one it actually is,” said Gelb.
FDNA seems aware of the software’s limitations, saying it can serve as “a reference tool” rather than replace medical tests.
“We showed that this system can be used in clinical settings,” Yaron Gurovich, chief technology officer at FDNA, told CNN.
No comments:
Post a Comment