AI-Driven Versus Nurse-Guided Expressive Writing for Cancer Patient Psychological Well-being: A Randomized Controlled Trial

This project provides access to the de-identified individual-level data associated with the research article entitled "AI-Driven Versus Nurse-Guided Expressive Writing for Cancer Patient Psychological Well-being: A Randomized Controlled Trial. The study investigated the effectiveness of an AI-driven expressive writing intervention compared to a nurse-guided intervention and usual care for improving psychological well-being in cancer patients. The dataset includes variables related to group assignment (AI-driven, nurse-guided, usual care), gender, education level, general residence area (using broad geographical categories), cancer type (with rare types categorized as "other"), cancer stage, current treatment status, data completion status, and scores on the Patient Health Questionnaire-9 (PHQ-9), the Generalized Anxiety Disorder 7-item Scale (GAD-7), the Fear of Progression Questionnaire-Short Form (FoPQ-SF), and the Connor-Davidson Resilience Scale (CD-RISC) at each assessment time point (baseline, post-intervention, and one-month follow-up). To ensure participant confidentiality, the following de-identification procedures were implemented: removal of all direct identifiers (e.g., names, ID numbers), generalization of geographic locations, and aggregation of rare cancer types into an "other" category. The research team has taken these steps to minimize the risk of re-identification while preserving the scientific utility of the data. The Southern Medical University Biomedical Ethics Committee has approved the sharing of these de-identified data. To enhance reproducibility and transparency, the library of writing prompts, the system-level prompts used to guide the AI chatbot, the training materials used for nurses, and the detailed manual for the nurse-guided intervention are available as supplementary materials with the published article.