Cochlear implant datalogging accurately characterizes children's 'auditory scenes'.

Area of Science:

• Audiology
• Speech-Language Pathology
• Pediatric Health

Background:

• Understanding children's auditory environments is crucial for speech development and language acquisition.
• Cochlear implant (CI) technology offers auditory rehabilitation, with processors often featuring automatic scene classification (ASC).
• Accurate environmental sound data is essential for evaluating CI effectiveness and identifying potential areas for technological improvement.

Purpose of the Study:

• To evaluate the accuracy of CI ASC algorithms in capturing children's real-world auditory environments.
• To quantify the duration and sources of electronic device use within these environments.
• To compare CI ASC classifications with human-coded auditory scene data.

Main Methods:

• Seven young children (mean age 36.71 months) with CIs participated.
• CI datalogs from Nucleus® 6 processors were analyzed alongside day-long audio recordings from a Language ENvironment Analysis (LENA) system.
• Auditory scenes were classified by both the CI algorithm and human coders for comparison.

Main Results:

• CI ASC showed strong correlation with human coding (ICC = .86) but with notable differences.
• ASC identified significantly more 'Quiet' scenes (p=.003), while human coders identified more 'Speech' (p=.04).
• Children were exposed to electronic sounds for an average of 8% of their day, primarily from mobile devices (39.7%).

Conclusions:

• CI ASC provides a generally accurate reflection of children's auditory environments.
• Discrepancies in 'Quiet' and 'Speech' classification highlight the importance of defining scene categories.
• The significant presence of electronic sounds suggests a need for a dedicated category in CI ASC systems.