Advertisement
Artificial intelligence
TechInnovation

Chinese scholars use AI to screen newborns for genetic disorders via facial scan

  • The new assistive diagnosis tool was designed to detect more than 100 disorders with distinctive facial features

Reading Time:2 minutes
Why you can trust SCMP
Researchers from the Shanghai Children’s Medical Centre and the Shanghai Paediatric Centre said their new AI-based assistive diagnosis tool would be used for initial screening, helping avoid missed or wrong diagnosis of newborns. Photo: Xinhua
Sarah Daiin Beijing

Chinese scholars have developed artificial intelligence-powered software that is geared to help screen newborns for genetic disorders through a facial scan.

Researchers from the Shanghai Children’s Medical Centre and the Shanghai Paediatric Centre said their new assistive diagnosis tool, which they described as the first of its kind, was designed to detect more than 100 disorders with distinctive facial features, including Cornelia de Lange syndrome (CdLS) and Down syndrome, according to a report by state-owned China News Service on Friday.

They said their AI-enabled tool would be used for initial screening, helping avoid missed or wrong diagnosis of newborns.

Advertisement

Prominent facial features of children with CdLS, for example, include thin eyebrows that often meet at the midline, long eyelashes, short upturned nose, thin downturned lips, low-set ears and high-arched palate or cleft palate, according to the CdLS Foundation. Children with Down syndrome may have tiny white spots in the coloured part of their eyes, flatter faces, a tongue that hangs out of the mouth and eyes shaped like almonds, or shaped in a way that is not typical for their ethnic group.

Powered by so-called deep learning algorithms, the new tool will be used to review past medical records for facial profile patterns and make use of that data to help diagnose new cases.

Advertisement
Advertisement
Select Voice
Choose your listening speed
Get through articles 2x faster
1.25x
250 WPM
Slow
Average
Fast
1.25x