Towards a detection and correction algorithm for expanding applicable time-range of EMAs for challenging situations

Authors: Anna Josefine Munch Sørensen1,2, Niels Henrik Pontoppidan1,2, Aleksandra Koprowska2, Lena Havtorn1,2, Soffi Skovlund Jensen1,2

1Eriksholm Research Centre
2Oticon A/S

Background: Hearing aid fittings play a crucial role in improving the quality of life for individuals with hearing impairments. To achieve optimal outcomes, it is essential to understand their challenging listening experiences in everyday contexts. However, when employing Ecological Momentary Assessment (EMA) as a tool to sample these experiences, researchers often encounter underrepresentation of challenging situations. MarkeTrak and other studies suggest that speech-in-noise scenarios remain the situations where hearing aid users more often express dissatisfaction with their hearing aids. Unfortunately, capturing these situations via EMA presents practical difficulties. For instance, recent papers concluded that participants found it socially inappropriate to extract a phone during a conversation. Consequently, we face the need to facilitate more accessible reporting mechanisms for users.

Method and Results: Analyzing data from a recent study, we observed drops in SPL to below realistic levels (Figure 1) before filling out EMAs, suggesting that participants physically distanced themselves from the evaluated situation. Consequently, when computing acoustic sound environment data around the time of EMA completion, the context may not precisely align with the moment of assessment. To address this limitation, we analyzed data to develop a detection and correction algorithm that could link the EMA response with the more appropriate objective sound environment data. Analyzing the situations from the recent study we found that communication situations led to more detections of environment change than non-communication situations (Figure 2).

Outlook: Moving forward we intend to validate the detection and correction algorithm by integrating it into the EMA app. By asking whether the previous situation is the one that was being rated, we would both validate the detection and correction algorithm, whilst also granting participants flexibility in their reporting timing.