gms | German Medical Science

87. Jahresversammlung der Deutschen Gesellschaft für Hals-Nasen-Ohren-Heilkunde, Kopf- und Hals-Chirurgie e. V.

Deutsche Gesellschaft für Hals-Nasen-Ohren-Heilkunde, Kopf- und Hals-Chirurgie e. V.

04.05. - 07.05.2016, Düsseldorf

Towards pre-surgical CI outcome prediction

Meeting Abstract

  • corresponding author Martin Billinger - Medizinische Hochschule Hannover, Hannover
  • Tina Peschel - Medizinische Hochschule Hannover, Hannover
  • Thomas Lenarz - Medizinische Hochschule Hannover, Hannover
  • Andreas Büchner - Medizinische Hochschule Hannover, Hannover

Deutsche Gesellschaft für Hals-Nasen-Ohren-Heilkunde, Kopf- und Hals-Chirurgie. 87. Jahresversammlung der Deutschen Gesellschaft für Hals-Nasen-Ohren-Heilkunde, Kopf- und Hals-Chirurgie. Düsseldorf, 04.-07.05.2016. Düsseldorf: German Medical Science GMS Publishing House; 2016. Doc16hnod272

doi: 10.3205/16hnod272, urn:nbn:de:0183-16hnod2720

Veröffentlicht: 30. März 2016

© 2016 Billinger et al.
Dieser Artikel ist ein Open-Access-Artikel und steht unter den Lizenzbedingungen der Creative Commons Attribution 4.0 License (Namensnennung). Lizenz-Angaben siehe http://creativecommons.org/licenses/by/4.0/.


Gliederung

Text

A cochlear implant (CI) restores hearing to the profoundly deaf. Although, being successful in general, speech comprehension performance varies greatly between individuals. This variability makes pre-surgical outcome prediction difficult. In order to get accurate predictions, possible sources of variability need to be identified. Although, the duration of auditory deprivation appears to play an important role, this factor alone does not allow accurate predictions. In our work we aim at identifying additional clinical factors and describing how they interact on final CI performance.

We analyzed data from 68 patients who received a CI. Their CI performance was measured as speech perception in noise (Oldenburger Satztest, OLSA) six months after implantation. In addition, 36 different factors were measured prior to the implantation. These factors cover a wide range of domains, including demographic data, auditory and visual text perception abilities, questionnaires, and history of hearing loss. We applied Random Forests to impute missing values, obtain relative factor importance, and to perform predictions.

The top four ranked factors are duration of hearing loss, age, duration of hearing aid use, and visual text reception. These factors were ranked significantly higher than a randomized dummy factor.

In summary, we successfully identified plausible factors for CI outcome prediction. These are first results, and the current analysis cannot reveal all possible factors. We expect to identify more factors in our continuing work, which will finally result in a predictive model for CI performance.

Der Erstautor gibt keinen Interessenkonflikt an.