A Prescriptive Design Framework for CRM-Integrated Agentic AI in Hospital Patient Services: Design Science Development and Empirical Proof-of-Concept Evaluation
Year: 2026
Type: article
Abstract: Hospital patient services (registration and appointment scheduling) often face fragmented systems, long handling times, and data-entry errors, while patients expect instant 24/7 digital interactions. This study uses Design Science Research to develop a prescriptive five-phase framework for CRM-integrated agentic AI, whose novelty lies in combining bounded transactional agency, middleware-mediated access to CRM/SIMRS tools, and staged governance. Its design assumptions are evaluated through a controlled proof of concept (PoC). The PoC implements an end-to-end transactional conversational workflow (WhatsApp/web chat) in which an LLM-based agent uses API middleware to call tools on a CRM/SIMRS sandbox, avoiding direct database access. The artefact structures are: preparation and alignment; analysis and detailed design; implementation and controlled testing; pilot and evaluation; and scaling and institutionalization, supported by governance mechanisms (clinical “green/red” boundaries, audit logging, and human-in-the-loop supervision). Empirical evaluation applied black-box testing across 15 positive and negative scenarios comprising 285 executions. Of these, 246 met the predefined scenario criteria, and 39 failed, yielding an overall controlled-scenario success rate of 86.32%. A separate message-level latency log contained 292 message-response pairs across 58 recorded conversation groups; median response time was 9 s (IQR 4–19 s), with 88.70% of responses returned in under 30 s and 97.26% in under 60 s. Selected persisted CRM records were additionally checked through authorized visual inspection, with no discrepancies observed. These findings support technical feasibility under controlled PoC conditions while identifying patient-centred evaluation, cybersecurity assessment, production load testing, economic evaluation, longitudinal monitoring, and multi-site validation as requirements for broader deployment.
Authors: Yohana Prihessy, Sfenrianto Sfenrianto
Institution: Binus University
Language: English
Subfield: Health Informatics
Field: Medicine
Domain: Health Sciences
Open Access status: diamond
OpenAlex ID: W7223176137