Investigating the benefit and relevance of the automatic classifier Autosense Sky in various complex acoustic conditions in CI children

More and more modern hearing devices are equipped with an automatic classifier,
which uses signal processing and machine learning techniques to categorize an
acoustic environment into one of several general categories. Based on the
selected category, different hearing device settings, such as the microphone
directivity, are activated. The requirements that such an automatic classifier
has to fulfil are different for children than for adults, because of the
different listening situations that children encounter. For example, children
are more often addressed from other directions than the frontal direction, when
they are playing or in the classroom. Whereas a directional microphone can
improve speech perception from the front in noisy situations, it also
attenuates sounds coming from other directions and should therefore only be
turned on in appropriate situations, especially for children with severe
hearing loss using a cochlear implant (CI). AutoSense Sky OS is an automatic
classifier designed especially for children and is used in Phonak Sky hearing
aids and Advanced Bionics Sky cochlear implants. This study tests the
hypothesis that AutoSense Sky OS improves speech perception from the front and
does not impair localization ability or sound identification from other
directions in CI children.

Gaining insight in the potential benefit and relevance of the automatic
classifier in various acoustic conditions for children using Advanced Bionics
Sky CI(s).

The study is a double-blinded crossover intervention study. Children test two
different settings of their CI(s) during the study: their current setting
(Autosense is turned off), and a new setting where Autosense is turned on. They
test the two settings (blindly) at home for two weeks each. For each of the two
take-home periods, they fill out a questionnaire and diary about their auditory
performance together with their parents. Furthermore, the children do listening
tests when they come back to the clinic, to compare their performance with each
of the two settings. These listening tests take place in a virtual classroom,
using a VR headset and loudspeaker ring. Speech recognition from the front is
tested using stimuli from the clinically-used digit-in-noise test. Furthermore,
sound detection/identification from other directions is tested, by letting the
children push a button when the hear an animal name (one button for tiger,
another button for rabbit). In these listening tests, a third condition is
included, which is the condition where a remote microphone is used by the
speaker from the front. Moreover, a technical evaluation is done to check how
Autosense reacts to the virtual test environment.

The automatic classifier Autosense Sky is turned on in the CI processor.

The participants need to visit the Erasmus MC twice, with four weeks in
between. After two weeks, the CI processor with the second program can be
picked up or send to their home address, and participants fill out a diary and
questionnaire with their parents to describe their experience in different
listening situations. During both visits, speech tests will be done, and each
visit will take 1-1.5 hours. After the second visit, the participants decide
together with their parents which program they want to keep. The sound of the
CI programs in the study will be very similar to the programs that the
participants are used to, so we expect no problems with acceptance of sound or
deterioration of daily auditory functioning during the study.