Layered Automation in Hospital Crisis Readiness: A Scoping Review and Framework for Deterministic, Intelligent, and Agentic Systems

Abstract

Hospitals face recurring crisis conditions, including pandemics, mass casualty events, natural hazards, and infrastructure disruptions, which stress operational coordination, compress decision cycles, and amplify information fragmentation. Automation has increasingly been used to support crisis responses through workflow execution, decision support, and predictive modelling. However, crisis automation literature remains fragmented across system types, hazards, and organizational contexts, and lacks a governance-aware synthesis that distinguishes between levels of autonomy and corresponding accountability requirements. This study reports an scoping review using EBSCOhost databases (PRISMA-ScR) of hospital crisis automation. Across CINAHL Ultimate and MEDLINE Ultimate, two search families (core automation; AI/agent terminology) yielded 2,297 records. A supplementary AI-assisted discovery search yielded 1,000 additional records for review. After screening predefined inclusion criteria and deduplication, 116 studies were included in the analysis. The synthesis identifies two dominant strata of crisis automation: deterministic workflow automation and bounded intelligent decision augmentation. In contrast, policy-constrained agentic automation, defined as goal-directed orchestration across heterogeneous systems, appears largely absent as a mature operational form in literature. Building on this synthesis, we propose a layered framework that links automation strata to integration patterns, failure modes, and escalating governance controls. A concrete use case illustrates how deterministic, intelligent, and agentic layers would differ in a surge scenario, highlighting practical constraints regarding data access, interoperability, and human authorization. This study contributes (i) a consolidated map of hospital crisis automation research, (ii) a governance-aware layered framework for disaster public health and healthcare informatics, and (iii) implications for designing and evaluating emerging agencies in crisis settings.