abstract
- Conformance and compliance checking of business processes help verify adherence to established rules, standards, internal policies, and ethical requirements. When deviations are detected, it is essential to communicate them effectively in ways that support corrective action. In this study, we focus on healthcare worker training¿specifically phlebotomy¿where procedural accuracy is critical for patient safety. Building on prior work that extracts structured execution traces from videos of trainees performing phlebotomy, we propose a conformance-to-feedback transformation approach for generating multimodal, role-aware feedback from diagnostic outputs. Our prototype delivers feedback in two complementary forms: (1) time-aligned visual guidance embedded in the trainee¿s recording and (2) structured natural-language reports tailored separately to trainees and instructors through constrained LLM prompting. A pilot evaluation with healthcare practitioners yielded positive results in terms of usability and perceived pedagogical value. The results suggest that multimodal, temporally grounded, and role-aware feedback can enhance procedural training by making automated conformance diagnostics more interpretable and actionable for non-expert users.