Artificial Intelligence Platform Architecture for Hospital Systems: Systematic Review.

Area of Science:

• Healthcare Informatics
• Artificial Intelligence in Medicine
• Digital Health

Background:

• Traditional hospital information systems hinder data-led decision-making due to data silos and fragmented workflows.
• Artificial intelligence (AI) platforms are crucial for revolutionizing healthcare, but their implementation faces challenges.
• Existing systems lack sufficient clinical intelligence for effective data utilization.

Purpose of the Study:

• To develop a 5-layer hospital-specific AI architecture (infrastructure, data, algorithm, application, security, and compliance).
• To systematically review and assess the applicability of existing evidence mapped onto this framework.
• To evaluate the maturity of different AI implementation layers in hospitals.

Main Methods:

• Systematic literature search across major databases (Web of Science, Embase, PubMed, Scopus) up to May 2025.
• Adherence to PRISMA guidelines with two independent reviewers screening studies.
• Inclusion of empirical studies on hospital-based AI implementations, excluding non-English and gray literature.
• Quality assessment using the Critical Appraisal Skills Programme tool and quantitative mapping using an ordinal maturity scale.

Main Results:

• 29 studies from 11 countries met eligibility criteria, covering diverse clinical domains.
• Application and data layers showed highest maturity (mean scores ~3.0-3.17), followed by algorithm and infrastructure.
• Security and compliance layer exhibited the lowest maturity (mean 1.69) and highest variability, with weak alignment to the technical core.
• Strong interconnections were found among data, algorithm, and application layers (Jaccard similarity 0.80-0.89).

Conclusions:

• The study validates a 5-layer hospital AI platform architecture based on empirical evidence.
• Clinical feasibility of AI in hospitals is achievable, but sustainability requires enhanced focus.
• Increased investment in infrastructure, data governance, and security/compliance is essential for widespread, sustainable hospital AI adoption.