Our Research Lab
Automated Hearing Aid Fitting (ISPS)
Basic hearing aid fitting is a complex but imperfect process. Conventional fittings derive the amplification "prescription" or "targets" (technically, the gain-frequency response model) from complex fitting formulae that are based largely on normative data from a variety of research studies that are linked to the individual patient only by their audiogram. As such, most patients have more amplification nor less amplification that is ideal or desired. In this project, we are working with our corporate partner Securboration and several expert scientist-clinicians, clinical scientists, and engineers to improve upon the traditional methods for establishing amplification (gain) settings for hearing aids and other amplification devices. The goal is to develop clinically viable and advantageous methods to derive an Individual Signal Processing Strategy (ISPS). We begin by deriving an individualized gain-frequency model but the method can be extended to any hearing aid parameter set.
Our strategy is simple and elegant. The chief complaint of someone with hearing loss is difficulty understanding speech in noisy environments and the primary complaint about hearing aid technology is there ability to help in those environments. So to obtain an individualized gain-frequency model (i.e., to fit the hearing aid to the individual), we engage the wearer in a speech-in-noise task. We have adopted a categorical perception task in background babble. This task it is easy to perform, the stimuli can be custom synthesized to meet the needs of frequency-dependent hearing aid fitting, and the process is extremely efficient. Efficiency is enhanced by using advanced computational methods such as machine learning to quickly converge on gain-frequency responses that maximize performance in the categorical perception task. This procedure accounts for and compensates for individual differences in speech perception and preferred speech audibility, particularly in background noise. Pilot data indicated that the ISPS method provides laboratory outcomes that are as good as the best-practices NAL-NL2 fitting-to-target method. Current work focuses on improving efficiency, collecting larger data sets, and refining the backend algorithms for convergence on the individualized gain-frequency model.
Conceptual diagram of the Individualized Signal Processing Strategy approach.
This method can be used in many possible implementation scenarios. In the context of individualized gain-frequency models, this method could be encapsulated into existing manufacturer fitting software for use by audiologists and hearing aid dispensers. The method could also be used for automated fitting when the traditional depending model is not available or desired. Importantly, the same basic methodology can be used to precisely fit advanced signal processing parameters to the individual device wearer. To date, we do not have sophisticated methods for setting those parameters.