Abstract B108: Exploring breast cancer survivorship: Computational topic modeling of written essays from chinese breast cancer survivors in an expressive writing intervention

Objective: Expressive Writing (EW) is an intervention designed to enhance health by facilitating emotional and cognitive processing through writing. A culturally tailored EW program has been shown to improve the quality of life among Chinese breast cancer survivors. However, little is known about the content of these essays and how they contribute to the observed benefits. This study aims to identify the recurring themes and concerns within the narratives of Chinese breast cancer survivors, explore cultural beliefs and experiential variations, and provide insights to guide future interventions and healthcare policies to better support their survivorship. Methods: The dataset underwent standard preprocessing steps in natural language processing, including text cleaning, tokenization, stop word removal, and stemming, without any domain-specific modifications. Using unsupervised structural topic modeling with the R package “STM” we determined the optimal number of topics and identified key themes from 246 essays written by participants randomized into two intervention groups. Results: As an unsupervised approach, we ran topic models incrementally, testing potential values of K (the number of topics) ranging from 3 to 15. The results showed greater redundancy in models with larger K. Based on topic solution statistics—including Held-out likelihood, Residuals, Semantic Coherence, and Lower Bound—we selected a 5-topic solution (K=5) as the optimal model. Using human interpretation of the most representative words provided in matrices such as Highest Probability, FREX, Lift, and Score, we identified and labeled the five topics that emerged from the dataset as follows: 1) Breast cancer diagnosis and treatment; 2) Balancing work and family responsibilities; 3) Hospitalization and surgical experiences; 4) Emotional struggles and social stigma; 5) Coping through faith and finding hope (the most prevalent topic). Conclusions: The findings highlight the need for culturally sensitive interventions to address the key themes identified in the narratives of this largely overlooked population. Recommended interventions include programs designed to tackle social stigma and support psychological resilience; resources to help survivors manage the dual demands of work and family responsibilities during survivorship; or faith- or community-based interventions to build stronger social networks and provide emotional support. Future research may engage human supervised thematic analysis or anthropological approaches on a qualitative level to uncover the nuanced interpretation within the cultural context Citation Format: Yusi Aveva Xu, Eunju Choi, Jace D. Pierce, Dongjie Wu, Qian Lu. Exploring breast cancer survivorship: Computational topic modeling of written essays from chinese breast cancer survivors in an expressive writing intervention [abstract]. In: Proceedings of the 18th AACR Conference on the Science of Cancer Health Disparities; 2025 Sep 18-21; Baltimore, MD. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2025;34(9 Suppl):Abstract nr B108.