PROJECT SUMMARY
Hearing loss is one of the most common age-related diseases. Untreated hearing loss is associated with in-
creased risks of negative social-emotional health outcomes including loneliness and depression. Why hearing
loss is associated with poor social-emotional health is unknown. The primary treatment for hearing loss is hear-
ing aids, but only 30% of adults with hearing loss use hearing aids. Whether hearing aids reduce the risk of
poor social-emotional health is equivocal, and if they do moderate the risk, how they do is unknown. In order to
improve the effectiveness of audiologic interventions in reducing the risk of poor social-emotional health, it is crit-
ical to understand why hearing loss is associated with poor social-emotional health, whether the mechanisms of
this association are hearing-related, and how hearing aid use moderates the risk of poor social-emotional health
outcomes. This study addresses these questions by introducing a theoretical framework of hearing-related be-
havior and novel methods to quantify it in the real world. Hearing-related behavior is proposed to comprise both
auditory lifestyle, or how acoustically demanding and diverse the soundscapes a listener encounters in daily life
are, and communication engagement, or how well a listener is able to engage in conversation within sound-
scapes in daily life. In the proposed study, these domains are measured using long-form audio recordings and
ecological momentary assessments collected from the daily lives of listeners with hearing loss who do and do not
use hearing aids. Then, the relationship between these domains and social-emotional health is determined. In
Aim 1, auditory lifestyle differences between the two groups is quantified by measuring auditory lifestyle demand
(proportions of different types of soundscapes) and diversity (predictability of soundscapes) for each participant.
Long-form audio recordings of daily life are analyzed using machine learning classification to precisely character-
ize the soundscapes of each listener. Sound recordings are also paired with ecological momentary assessment
to add context. Whether hearing aid users spend more time in demanding (e.g., speech-in-noise, group conversa-
tion, unfamiliar talkers) and diverse (less predictable) soundscapes than non-hearing aid users is determined. In
Aim 2, recordings from the real world are identified where participants engaged in conversation. Conversations
are analyzed for utterance length and conversational turn rate, two measures of communication engagement.
Differences in engagement between hearing aid users and non-hearing aid users are determined. In Aim 3,
participants complete standardized and momentary measures of social-emotional health (depression, loneliness,
quality of life). The association between hearing-related behavior, hearing aid use, and social-emotional health is
determined. Findings from this study will determine how social-emotional health and hearing loss are related and
how clinical audiology can best promote positive behavior change that reduces the risks of loneliness and de-
pression in people with hearing loss. This study addresses goals of the NIDCD Strategic Plan by using machine
learning to develop new outcome measures that will improve clinical practice and intervention effectiveness.
