Designing Personalized Feedback Comments for Safety Management Based on Personality and Engagement: The PersonaTrace Scope Framework

Abstract

In high-risk work systems, safety is increasingly sustained not by procedural compliance alone but by operators’ continuous interpretation of unfolding situations, attention allocation, and adaptive coordination under time pressure. Yet operationalizing personalized safety interventions remains challenging due to (i) poorly specified personalization dimensions, (ii) unstable message quality at scale, and (iii) limited deployability in real organizations.We propose PersonaTrace Scope (PTS), a two-axis design framework that specifies what to say, to whom, and at what cognitive depth. PTS independently controls (a) message framing and tone through seven practice-oriented personality factors and (b) intervention depth—defined as behavioral demand and expected cognitive load—through five engagement levels. This yields 175 predefined scopes with corresponding structured comment templates.To sustain minimum quality across this corpus without exhaustive expert review, we use a large language model (LLM) strictly as a quality assessor—not as a content generator—and introduce the Behavioral Trigger Index (BTI), a multidimensional gate that quantifies action-triggering properties relevant to human reliability (self-efficacy, adaptability, immediate executability, self-relevance, and organizational influence). Comments are iteratively refined until they satisfy a prespecified BTI threshold.We conducted a field evaluation with approximately 150 nurses, assessing perceived agreement, usefulness for patient safety activities, and intention to apply the content during work. Positive responses (Agree/Strongly agree) were 80%, 72%, and 74%, respectively, indicating that PTS-delivered comments are acceptable and may support safety-related reflection and behavioral intention. PTS offers an operational pathway to reconcile personalization with scalable quality assurance in high-risk domains.

Keywords: Personalization, Safety Adaptability, Human Reliability, Engagement, Feedback

DOI: 10.54941/ahfe1007555