Repeated computational phenotyping of adaptive learning in chronic tinnitus: A pilot ecological momentary assessment study
Lili Zhang, Greta Monacelli, Himanshu Vashisht, Winfried Schlee, Berthold Langguth, Willie Muehlhausen, Tomás E WardObjective
Adaptive learning allows individuals to adjust their behaviour in response to changing environments and may be sensitive to fluctuations in tinnitus-related distress. However, it remains unclear whether momentary changes in tinnitus symptoms are accompanied by corresponding changes in adaptive learning within individuals. Although ecological momentary assessment (EMA) enables the investigation of symptom dynamics in everyday settings, repeatedly assessing computational phenotypes presents important methodological challenges. This pilot study examined the feasibility of integrating EMA with a repeated aversive learning task and explored whether momentary tinnitus distress was associated with within-person variation in behavioural and computational measures of adaptive learning.
Methods
Participants with chronic tinnitus completed baseline assessments followed by a two-week EMA protocol delivered through the AthenaCX mobile phone application. Momentary tinnitus loudness, tinnitus stressfulness, emotional valence, and concentration were sampled three times daily. A web-based Aversive Learning Task (ALT) was administered when tinnitus stressfulness reached predefined high or low thresholds. Behavioural and computational measures were analysed using mixed-effects models to examine within- and between-person variation.
Results
Ten participants were recruited and eight were included in the analyses. Participants completed 65.16% of all possible EMA prompts. Tinnitus loudness and tinnitus stressfulness showed substantial within-person variation and were positively associated. Several behavioural and computational measures also exhibited meaningful within-person variability across repeated assessments. Exploratory mixed-effects analyses suggested that higher momentary tinnitus stressfulness was associated with reduced adaptation of learning rate to environmental volatility.
Conclusions
This pilot study demonstrates the feasibility of combining smartphone-based EMA with repeated computational assessment of adaptive learning in individuals with chronic tinnitus. The findings provide preliminary evidence that some computational phenotypes may vary within individuals over time and may be associated with momentary tinnitus-related distress. Larger studies are needed to establish the reliability and generalizability of these effects and to determine whether EMA-based computational phenotyping can support future individualized assessment and symptom monitoring in clinical settings.