Introduction: Recovery is crucial for individuals with mental health challenges and is typically evaluated qualitatively. While the Recovery Assessment Scale (RAS) offers a quantitative measure, its cross-sectional nature limits its ability to capture the recovery process quantitatively.
Objectives: This study aims to quantitatively clarify emotional changes in the recovery process of individuals receiving community-based home-visit services by analyzing their occupational narratives using natural language processing.
Methods: The participants were users of psychiatric home-visit nursing services. Demographic and questionnaire data (including RAS) were collected, and the Occupational Performance History Interview-II (OPHI-II) was conducted, classifying interview content into three categories: past, present, and future. Sentiment scores (positive and negative emotions) were assigned using J-LIWC2015 (Igarashi, 2022). Statistical analysis included repeated measures ANOVA (RM-ANOVA) to compare the three categories and a generalized linear mixed model (GLM) to analyze trends based on median RAS scores. This study was approved by the ethics committee of the institution to which the authors belong.
Results: The study included 24 participants. The RM-ANOVA revealed significant changes in negative emotional word frequency across time categories (p < 0.001), with a notable increase in future-related narratives (vs. past: p = 0.005, vs. present: p = 0.011). The GLM showed these changes were significantly larger in the lower RAS score group (p = 0.005). These findings suggest that narrative emotions fluctuate over time, and individuals with lower RAS scores tend to use more negative emotional words when narrating about the future.
Conclusion: The results suggest that the recovery process can be quantitatively measured over time using natural language processing and highlight the importance of future-related narratives in recovery.