AI could help predict language development in kids with cochlear implants

Embargoed until: Publicly released: 2025-12-27 03:00

Australia; International; VIC

Photo by Laura Ohlman on Unsplash. Story by Olivia Henry, Australian Science Media Centre

Aussie and international researchers may have found a way to better predict spoken language development in kids with cochlear implants. The study looked at 278 kids with cochlear implants who spoke English, Spanish, or Cantonese, and say applying a deep transfer learning model to their brain scans, a type of AI which uses prior knowledge learnt from pretraining on a large dataset, achieved 92.39% accuracy in predicting spoken language improvement – which they say is superior to conventional approaches. Their findings could help predict whether kids will have high or low spoken language improvement after a cochlear implant, the team says, and is a first step towards creating customised treatment plans for kids who need cochlear implants.

Attachments

Note: Not all attachments are visible to the general public. Research URLs will go live after the embargo ends.

Research JAMA, Web page The URL will go live after the embargo lifts.

Journal/
conference: JAMA Otolaryngology–Head & Neck Surgery

Research:Paper

Organisation/s: The University of Melbourne, Lurie Children’s Hospital of Chicago, USA

Funder: This work was supported by the Research Grants Council of Hong Kong (grant GRF14605119), and US National Institutes of Health (grants R21DC016069 and R01DC019387). Conflict of Interest Disclosures: DrWong reported grants from the US National Institute on Deafness and Other Communication Disorders (R01DC019387) and the Research Grants Council of Hong Kong (GRF14605119) during the conduct of the study, as well as nonfinancial support from Foresight Language and Learning Solutions Ltd and a patent for US11607309B2 issued outside the submitted work. Dr Young reported grants from the US National Institute on Deafness and Other Communication Disorders (R01DC019387) during the conduct of the study as well as grants and personal fees from MEDEL and a patent for US11607309B2 issued outside the submitted work. No other disclosures were reported.

Media Contact/s

Contact details are only visible to registered journalists.