Genetic algorithms
 
Cochlear (US) performed a study where they used a genetic optimisation algorithm to optimise map parameters (Lineweaver et al., 2004). T and C levels were not included in this method but a ‘T-level modifier’ of 20% was included. Furthermore it included things like rate, number of maxima, Q-level, etc. Genetic optimisation also includes spontaneous mutations.
Genetic optimisation means that you select a subset of parameters from a set that give the best results and than create descendents that are a combination of the parameters that were selected. This process is designed to mimic a natural evolution process (survival of the fittest).
In practice, in the study, the recipients needed to select 4 maps from a group of 8. The software then creates 8 new maps and the process repeats.
In the first pilot experiment 5 experienced subjects were tested. All 5 converged to a map that was different from their original map but had no significantly different speech score. Convergence took up to 20 generations.
The fact that users converge to a map different from their own is a promising outcome, since generally recipients prefer the map that they are accustomed to. This means that in theory they have the ability to find a map that is not just the default.
A second study is ongoing.
 
Back to Cochlear Implant fitting.