Task T9.1 · Gantt: Conduct readiness assessments across participating NPHIs

Data Readiness Assessment in NPHIs

This task conducts readiness assessments across participating NPHIs to establish the infrastructure, governance and capacity baseline relevant for subsequent deployment.

Key activities

· Collect and structure the evidence, partner inputs or consultation results relevant to the task scope
· Analyse the findings and identify the implications for implementation, quality assurance or alignment
· Consolidate the outputs into a validated basis for project delivery, decision-making and reporting

You can browse the questionnaire freely, but submitting requires a beneficiary or administrator account. Sign in before filling in your answers.

The grey hints under each question show example answers from a completed card, to illustrate the expected level of detail. Feel free to skip questions outside the scope of your collaboration.

Respondent information

Full name *

Email *

NPHI / Institution *

Country *

Role / function

Implementation plan — Data infrastructure readiness in NPHIs

(1) Is there in the country at least 3 health facilities with EHRs?

(2) Are other data sources such as registries accessible electronically? (e.g., DHIS2, national registries...)

(3) Is there a central server that can be used for the project?

1. EHR Data Structure & Standardization

How is your EHR data currently structured (proprietary, internal standard)? Is it in a standard representation format with coded terms? Are there clinical notes as unstructured text?

Optional: specify the vendor if off-the-shelf. Is there an analytical data layer (Data Lake / Warehouse)?

Is your data already mapped to a recognized standard (e.g., OMOP, FHIR)? If so, which one and how broad is the mapping? If not, is a data dictionary available?

Have you previously collaborated with third parties to convert your EHR data into OMOP or FHIR? Which technology was used?

Have you participated in projects requiring federated access to your EHR information? If yes: (a) technical specs of your federated node, (b) platform/software used, (c) typical Data Access Committee turnaround.

2. Available Data Types & Sources

What care levels do you have access to: Primary Care, Secondary Care, Urgency/Emergency? Do you integrate home-monitoring results?

Which clinical domains are accessible: Mortality, Diagnostics (ICD10/SNOMED), Medication (ATC), Hospitalizations, Lab results (LOINC), Vital signs, Symptoms, Procedures? Any NLP solution for unstructured data?

Do you have access to structured Biomarker Data (e.g., blood/urine/spirometry tests, anthropometric data)? Please specify.

What is the typical update frequency for operational/prospective data (e.g., patient admissions, resource status): daily, hourly, continuous feed?

What is the current circuit/infrastructure for sending personalized recommendations to patients, if any?

3. Third-Party Application Integration

Have you previously worked with and successfully integrated third-party applications within your EHR/hospital system?

What are the key technical and regulatory requirements for integrating deployed outputs (e.g., risk metrics, suggestions) into your existing workflows?

What infrastructure is currently used for receiving and displaying these recommendations/outputs?

4. Resource Optimization Needs & Practices

How are decisions currently made regarding ambulance placement, bed reservation and staff management?

Do you currently use any forecasting, predictive or optimization tools to inform these decisions? If so, please specify.

Have you put any predictive solution (e.g., predicting hospitalization risk) into a pilot program or current practice?

What are your system's most pressing resource optimization needs right now?

What is the usual process and duration for requesting access to retrospective EHR data, including the full governance timeline (ethical committee clearance) before a longitudinal study?

What are the computational characteristics of your centres (CPU/GPU, storage, network bandwidth) so we can anticipate additional resource needs ahead of federated training?

5. Individual-Level Data Linkage & Governance

Common identifiers: is there a unique national identifier available across datasets to allow deterministic linkage of records at the individual level?

Legal and ethical framework: are there existing permissions to link these data sources for research, or must new ethical approvals be secured?

6. ETL Pipelines, Restricted Repositories & Federated Analytics

Is it allowed to deploy a project-specific ETL pipeline at your site to map source data into OMOP/FHIR, provided it runs entirely within your local infrastructure and no raw data leaves the site?

Is it acceptable to create a dedicated, access-restricted data repository containing only the subset of data required for the planned analyses (data minimisation)?

Can such a restricted repository be technically and logically separated from operational clinical systems while remaining within the same institutional security perimeter?

Is it permissible to deploy federated analytics or federated learning agents (e.g., DataSHIELD or similar) that execute locally and return only aggregated, non-disclosive results?

Are there internal governance processes, security reviews or data protection impact assessments required before deploying ETL pipelines, restricted repositories or federated analytics components?

Are there preferred platforms, tools or architectural patterns that your organisation requires or prohibits for ETL, data storage or federated analysis deployments?

7. Prospective Validation & Test Environments

Does your organisation have access to non-production environments (EHR test systems, training systems, synthetic/anonymised repositories) usable for demonstration and validation of clinical decision support services?

Within a test environment, could data be retrieved for a limited, predefined set of pilot patients (synthetic patients or test accounts) for prospective validation and usability testing?

Is it possible to provide controlled access to a selected group of pilot users (clinicians, policy makers) for usability evaluation of CDS services in a test environment?

Are there preferred standards, integration mechanisms or security requirements for deploying CDS services in test environments at your site?

Is there a test climate-data repository that could be used for pilot studies?

8. Source Database Profiling & Local Server Specifications

Could you run the OHDSI WhiteRabbit scan (github.com/OHDSI/WhiteRabbit) on the in-scope source data and share the scan report? Are data dictionaries or source descriptions available?

Processing power (CPU/GPU): what processing resources are available, particularly dedicated GPUs for model training?

Memory (RAM): specifications of the local servers for extraction and processing of high-volume EHR data.

Storage: available disk space and storage type for hosting the standardized databases (OMOP/FHIR) and supporting federated training.

Network and connectivity: describe the network environment hosting the server (restricted clinical network, DMZ) and supported security protocols for data exchange.

Software environment: is containerization (e.g., Docker) supported to facilitate deployment of federated learning and data standardization tools?