AI model for predicting adult cochlear implant candidacy using routine behavioral audiometry

Area of Science:

• Audiology
• Artificial Intelligence
• Medical Informatics

Background:

• Cochlear implantation is a complex decision with varying candidacy criteria.
• Predictive models can aid in identifying suitable candidates for cochlear implants.

Purpose of the Study:

• To develop and describe an artificial intelligence (AI) model for predicting adult cochlear implant candidacy.
• To utilize basic demographic and standard behavioral audiometry data for prediction.

Main Methods:

• A machine-learning approach was employed using retrospective data from adults undergoing cochlear implant candidacy testing.
• The model predicted CNC word scores and AzBio sentence in quiet scores.
• Performance was compared against published prediction tools and benchmarks.

Main Results:

• The model, trained on 770 adults, showed isophoneme and word recognition scores as key predictors.
• Mean absolute differences were 15% for AzBio and 13% for CNC scores.
• The combined model achieved 87% accuracy (90% sensitivity, 80% precision).

Conclusions:

• An adaptive AI model can predict adult cochlear implant candidacy using routine audiometric and demographic data.
• Implementation includes a public online tool, a smartphone app, and EHR integration.
• The model facilitates identifying and engaging potential cochlear implant candidates.