 I'm Julia Agoneze and I welcome you to my talk towards an objective measurement of individual listening preferences, trade consistency and state specificity. Proper hearing aid fitting is crucial for users' satisfaction and listening comfort and this fitting process includes a first prescriptive state where information on the hearing threshold is collected and a fine-tuning stage involving measures at super-threshold levels which often follows a process of trials and errors where the technician adjusts the settings according to the patient's complaints and needs. Despite being very time-consuming, this stage is crucial for user satisfaction and needs good individualization and personalization. Indeed, for proper fine-tuning, individual listening preferences should be considered. Folk and colleagues assume that individuals on preference judgments are based on different subjective traits, with individuals showing different preferences along three trade-offs, intelligibility versus listening comfort, amplification preferences, and noise distortion preferences. On this, we will now avoid further. The use of single-microphone noise reduction algorithms in hearing aids introduces an important trade-off between noise and distortion. These algorithms provide an increase in signal-to-noise ratio by reducing the surrounding noise. But with these, they also introduce distortions to the signal. Here we can hear an example of a clean signal of the same signal with background noise and with distortions. So these noise reduction algorithms are however an essential part of the hearing aid fitting process. Since understanding speech and noise is one of the main challenges for people with mild to moderate hearing loss, but the problem is that the present noise measurement is available to determine noise reduction performance. So how can we measure this trade-off preferences for better individualized fitting? Previous studies have shown that preference on acceptable noise level predicts pattern of hearing aids used with 85% accuracy and that there is considerable inter-individual variability in preferred noise reduction settings for experienced hearing aid users. And these preferences were found to be stable after one year. And with respect to the noise distortion trade-off, a measure to estimate this individual trade was first conceptualized by Polka and Kulnik and later implemented by Kuvja, who in our study with both normal hearing and hearing impaired participants identified three individual profiles. Noise haters, distortion haters and intermediate. In addition, she also found stability of these preferences over time in a test-free day after one week and over listening scenarios, namely with different maskers and special conditions. With respect to potential predictors of these individual preferences, acceptable noise level was not seen to be associated with self-reports while a relationship between noise reduction preference and putain average was found with higher PTA for individuals who prefer strong noise reduction. Finally, noise distortion preferences have been shown to correlate with speech reception pressure. So we can here conclude that noise distortion preferences have been found to be an individual trade stable over time and modernities. However, time stability was only measured at test-free tests over one week and one year, but how about potential day-to-day fluctuations? Indeed, we know that other new psychological processes like cognition and hearing itself fluctuate over time within the individual. Moreover, up to now these measurements of hearing preferences happened in a highly controlled experimental setting, but how about an ecological mobile assessment in the everyday life of the individual? Therefore, the aim of our study is to move towards an objective measurement of listening preferences that can be administered in a mobile app. We will assess a newly implemented mobile task and evaluate its psychometric quality and use this measure to evaluate stability of listening preferences along different days within the individual. For this, we collected data from 185 unedited individuals who reported subjective hearing difficulties. The study was conducted along three weeks, was entirely online on the mobile phone of the participants. In week one, the study included the assessment of different questionnaires like demographics and preferences, personality, as well as the newly implemented noise distortion trade-off task. This task was repeated three times and preceded by a calibration and practice trial. The ecological momentary assessment measures that followed were spread along two weeks in the morning and evening. Here, participants were prompt by SMS to complete the assessment of noise distortion task once again, hearing performance and note. These noise distortion tasks assess individual preferences in three listening conditions, each of them displayed in the form of a slider with 90-square values. The participant was asked to move the cursor of the slider from left to right as much as needed to understand the speech with little effort. The three conditions and three sliders that made each task were a simple linear game scenario to assess the comfortable SNR level, a trade-off scenario with game at the expense of clipping distortions with a general SNR range, and again a trade-off scenario with an adaptive SNR range this time based on the individual answer on the first slider to evaluate the subject-dependent trade-off. To classify noise and distortion data, two measures were taken into account, the data SNR between condition or slider 1 and 2, so linear game versus general trade-off, and the data SNR between slider 1 and 3, so linear game versus subject-dependent trade-off. Noise haters would show a data SNR close to zero, meaning that the desire to maintain listening comfort, so no noise at the expense of signal quality, while distortion haters would show a positive data SNR, meaning that they accept lower listening comfort, so higher noise compared to the sweet spot to maintain better signal quality to avoid distortions. And the following research questions will guide our analysis. First, we focus on the segmented quality of the measurements and evaluated if measurements are consistent for different slider differences and which of the two data was the measure of choice. So both measures, namely data SNR between condition 1 and 2 and condition 1 and 3, are measuring the same individual rank order, and the data SNR between condition 1 and 3, so between linear game scenario and subject-dependent trade-off, is the measure of choice, since it shows less within person variability. Then having selected the most appropriate measure, we will evaluate traceability and state-related variants, and finally, we will see if a data-driven categorization of noise distortion haters is possible. First, we will apply a latent state-and-trade multi-level model framework with an estimation on autoregressive effects of each time point on the following measurement. And then we will explore associations between state variants and in mode, and between trade stability and self-reports of some purposes. Finally, a mixture model will be used for a data-driven identification of classes. The analysis of data at the moment is still ongoing, so we can't yet answer these last questions. However, from some preliminary data visualization, when looking at the overall sample distribution at baseline, we might already identify some individuals who are more noise haters versus distortion haters. And we can already see how some participants show quite clear preferences. Here, for example, we have a potential noise hater who showed similar preferences along the different logic-to-dial measurements, here represented by trial angles, and participants with way more variability. For now, I thank you for listening, and I will be happy to receive any question during the conference or by any mission.