Table 2 Comparison of various platforms and frameworks designed for the ingestion and use of RWD

Architecture

Hybrid-cloud with mandatory on-premises pseudonymisation.

Centralised, secure cloud enclave.

Typically, on-premises, with federated query capabilities.

Cloud-based, focused on automated data integration.

Open-source Python framework for EHR data analysis.

Data Privacy Model

Privacy by Design’: pseudonymisation occurs before data leaves the hospital environment.

Centralised de-identified patient data.

Data remains at local institutions; federated queries on aggregate counts.

Automated de-identification and extraction pipelines.

Software library; infrastructure-agnostic.

Collaboration Model

Dual support for centralised analysis and federated learning (NVFlare).

Centralised analysis within the secure enclave.

Federated queries and local analysis.

Centralised analysis of integrated data.

User-implemented; not an inherent feature.

Core Ecosystem

Deeply integrated into a single vendor ecosystem (Microsoft Azure).

Mix of tools within the NCATS secure environment.

Open-source ecosystem with multiple plugins and extensions.

Custom-built, disease-specific (oncology).

N/A (software library).

Regulatory Alignment

Proactively designed to align with ISO 42001:2023 and the EU AI Act.

Focused on data use agreements and IRB oversight.

Primarily focused on data standardisation (e.g., OMOP).

Focus on research-grade data curation for clinical outcomes.

N/A (software library).

Common data model (supported ontologies)

FHIR (UMLS)

Custom (N/A)

Custom (N/A)

N/A

N/A