A groundbreaking international study published in JAMA Otolaryngology-Head and Neck Surgery has shown that artificial intelligence (AI) can predict spoken language outcomes for children.
The AI model, which uses deep transfer learning, predicted spoken language outcomes one to three years after cochlear implants with an accuracy of 92%.
Cochlear implants: the only effective treatment for severe to profound hearing loss
Cochlear implantation is the only effective intervention to improve hearing and enable spoken language for children with severe hearing problems; however, spoken language development after early implantation is more variable than that of children with normal hearing.
Researchers trained AI models to analyze results based on pre-implantation MRI scans of the brains of 278 children in Hong Kong, Australia and the US who spoke three languages (English, Spanish and Cantonese).
Such complex, diverse datasets have been observed to pose problems for conventional machine learning; the study showed remarkable results with the deep learning model.
In this regard, senior author Nancy M. Young, MD, medical director, said: “Our results support the feasibility of a single AI model as a robust prognostic tool for the language outcomes of children using cochlear implant programs worldwide.”
The researchers believed that the same methods could eventually be used to predict success for pediatric conditions other than hearing loss.
Nevertheless, the pediatric cochlear implant program is one of the largest and most experienced in the world, with more than 2,000 cochlear implant procedures performed since its inception in 1991.
