Compliance-First Agentic AI Architecture Proposed for Hospital Systems
A new study released on arXiv (2608.06112) presents a compliance-focused, multi-tiered Agentic AI framework specifically designed for hospitals. This initiative aims to address the high failure rate, with 70-80% of healthcare AI projects not succeeding in scaling. The proposed framework upgrades existing hospital AI systems by adding an Agent Orchestration Layer for managing workflows across clinical, operational, and financial areas, along with a Compliance and Policy Layer that integrates policy-as-code for regulations like HIPAA, GDPR, the EU AI Act, DISHA Act, India’s DPDP Act, and ISO/IEC standards. The research highlights that many hospital AI efforts are siloed, leading to redundancy and unrecognized risks, and advocates for a more cohesive AI strategy to improve scalability and compliance.
Key facts
- Paper on arXiv:2608.06112 proposes a compliance-first Agentic AI architecture for hospitals.
- 70-80% of healthcare AI pilots fail to scale.
- Architecture includes an Agent Orchestration Layer for multi-agent workflows.
- Compliance and Policy Layer centralizes policy-as-code for HIPAA, GDPR, EU AI Act, DISHA Act, India's DPDP Act, and ISO/IEC standards.
- Most hospital AI deployments are isolated point solutions in departmental silos.
- The architecture extends existing hospital AI platform models.
- The paper addresses governance gaps, fragmented data, and missing integration blueprints.
- The research is announced as a new arXiv paper.
Entities
Institutions
- arXiv