JMIR Publications

ABSTRACT

Background:

Workday happiness is associated with workplace performance and burnout, but frequent questionnaire-based assessment is burdensome in real-world workplace settings. Free-text daily reports may provide a lower-burden way to monitor day-to-day changes in workday happiness.

Objective:

This study aimed to examine whether daily diary text can be used to estimate longitudinal within-person changes in workday happiness and to explore text-related factors, such as average sentence length and lexical diversity, associated with model performance.

Methods:

We collected free-text daily reports and self-reported workday happiness scores from employees in 2 Japanese companies. Company A provided training data from 94 participants over 2 months (1728 reports), and Company B provided test data from 11 participants over 6 months (652 reports). We used 2 text-based approaches: Bidirectional Encoder Representations from Transformers (BERT)-based regression model trained on Company A data, and a locally deployed Japanese large language model used in a zero-shot setting. Model performance was evaluated for each participant using the Pearson correlation coefficient between self-reported and estimated workday happiness scores, with r=0.40 used as a pragmatic feasibility benchmark.

Results:

9 of 11 participants met or exceeded the feasibility benchmark of r=0.40. Significant positive correlations were observed for 9 participants; for 1 participant, the correlation coefficient could not be calculated because the self-reported score remained constant, and 1 participant showed a small negative, nonsignificant correlation. Participant-level correlations ranged from -0.06 to 0.65 for the BERT model and from -0.05 to 0.81 for the local large language model. Performance appeared to vary according to diary characteristics such as sentence length and lexical diversity.

Conclusions:

Our study suggests that daily report text supports low-burden, longitudinal estimation of individual-level workday happiness. However, performance varied across participants, and further work is needed to improve generalizability, reduce attrition, address possible measurement bias, and clarify appropriate workplace use.