AI-enabled three-layer method to strengthen cancer patient safety plans at specialized centers in Peru

Aldana C.; Vilcahuaman L.; Vila K. · IEEE International Symposium on Medical Measurements and Applications, MeMeA (2026) · DOI: 10.1109/memea69746.2026.11537324

Resumen

Peruvian oncology referral centers use patient safety plans, safety walkrounds, incident reporting and indicators, but operational gaps remain in taxonomy consistency, signal integration, prioritization and timely closure of corrective actions. Objective: To propose a three-layer, AIenabled method that strengthens oncology patient-safety planning and execution across sites under human-in-the-loop governance. Methods: Layer 1 transforms institutional plans and annexes into computable management structures and a gap register, enabling auditable accountability through explicit owners, timelines, indicators, denominators and closure evidence. Layer 2 consolidates incident reports, walkround findings and indicator series into a dynamic risk profile, standardizing classification and prioritization to detect recurrent, high-impact risks by service and process stage. Layer 3 converts prioritized risks and plan gaps into measurable actions managed through workflow governance, tracking status and verifying closure with documentary evidence. The Patient Safety Committee validates classifications, prioritization and proposed actions, ensuring transparency, regulatory alignment and prevention of uncontrolled automation. The method closes the operational gap between written plans, observed safety signals and measurable actions, enabling cross-site comparability and KPI-driven monitoring of recurrence reduction and closure performance. The approach supports cross-site benchmarking and sustained organizational learning

Instituciones: Pontificia Universidad Católica del Perú