Modeling Population Psychometric Characteristics of a Speech-in-Noise Task: Using a Large Cross-Sectional Study to Explore Associations Between Cognition and Listening

• Activity Number: 355
• Type: Contributed
• Date/Time: Tuesday, August 5, 2014 : 11:35 AM to 12:20 PM
• Sponsor: Section on Statistics in Epidemiology
• Author(s): Mark Edmondson-Jones*+
• Companies: NIHR
• Keywords: Psychometric function ; estimation ; MCMC ; cognition ; hearing

Abstract

26% of adults aged 40 report difficulty understanding speech in background noise increasing to 48% by age 70. We investigate the characteristics of this decline in hearing function and its association with cognition.

UK Biobank (http://www.ukbiobank.ac.uk/) is a cohort study of over 500,000 UK resident adults aged 40 to 70 recruited between 2006 and 2010. A subgroup of c. 40k performed a speech-in-noise (SiN) hearing test and five cognitive tests. The adaptive SiN test involved listening, with each ear, to 15 spoken digit triplets played against a noisy background and keying them via a touchscreen. Instead of estimating a speech reception threshold (SRT) based on the mean signal to noise ratio (SNR) of the final triplets we estimate the SNR at which 50% performance is expected by modelling the underlying psychometric function, considering the effect of stimulus independent lapses. This was implemented using a Bayesian generalised linear mixed model framework. The model results describe how aspects of listening attention and performance vary with socio-demographic and cognitive variables.