Spontaneous emotional self-reports in digital mental health platforms: behavioural signatures and implications for affective regulation

Introduction

Spontaneous emotional self-reporting in digital mental health platforms remains poorly understood, despite its potential to capture ecologically valid aspects of affective experience. While prompted assessments have been widely used in ecological momentary assessment paradigms, unprompted emotional inputs may reflect distinct behavioural and psychological processes that are not captured by structured sampling approaches, particularly those related to intrinsic motivation, self-awareness, and context-dependent emotional salience.

Objectives

This study aims to characterize spontaneous emotional self-reports in a digital mental health platform by analyzing their emotional, temporal, and behavioral features, to explore their potential as markers of user engagement and affective processes; and to operationally define and preliminarily validate the concept of spontaneity based on temporal patterns of reporting

Methods

We conducted a retrospective observational study including 3,792 users of the eB2 MindCare platform who recorded at least one emotional self-report. Spontaneous reports were defined as unprompted entries occurring after a minimum of three days without prior reporting activity. Emotional (valence), temporal (frequency, timing, and inter-report intervals), and behavioural (variability, irregularity, and clustering) features were analysed at both individual and population levels. Distributions were summarised using descriptive statistics, and skewness and dispersion metrics were used to characterise temporal dynamics and inter-individual variability. Diurnal patterns were analysed based on time-of-day distributions.

Results

A total of 112,430 emotional self-reports were recorded across 323,550 user-days, of which 10,243 days (14.3%) included spontaneous reports. The median inter-report interval was 3 days, whereas the mean interval was 20.5 days, indicating a highly skewed distribution characterised by frequent short gaps and occasional prolonged inactivity. Emotional valence was predominantly negative, with 48.6% of users showing a negative median valence, although neutral and positive entries were also common. Reporting behaviour exhibited substantial inter-individual heterogeneity in frequency, temporal clustering, and emotional intensity. Spontaneous reports showed a clear diurnal pattern, with increased frequency in the evening and a peak around 19:30. Users with higher spontaneity displayed more irregular and episodic reporting patterns, with longer inter-report intervals but sustained participation over time. Behavioural metrics showed limited ability to distinguish between clinical cohorts

Discussion

Spontaneous emotional self-reports appear to represent a distinct behavioural phenomenon rather than a mere absence of prompted data, likely reflecting internally salient and subjectively meaningful affective states. Their irregular, event-driven nature and marked inter-individual variability suggest potential relevance as behavioural proxies of affective regulation processes. However, these reports likely represent a filtered subset of emotional experience, influenced by individual reporting tendencies and possible avoidance of highly distressing states. These findings highlight the need to reconsider how engagement is conceptualised in digital mental health, moving beyond response frequency towards longitudinal and context-sensitive patterns of interaction. Integrating spontaneous and prompted data may enable more ecologically valid and personalised approaches to monitoring and intervention in mental health, and it also may inform adaptive, user-centred digital phenotyping frameworks for future research and clinical applications.

References

No references