PHIRI APP - WP7 - PHIRI Federated Research Infrastructure (FRI) - D7.2 Mid-size prototype of PHIRI federated research infrastructure

D7.2 Mid-size prototype of PHIRI federated research infrastructure (PHIRI app)

The PHIRI Federated Research Infrastructure (FRI) is supported by a containerized reproducible solution for data analysis deployed by each participant partner on-premises.

This solution is based on identifying the relevant data sources for each case study (including the demonstration pilot), developing the common data models and analytical pipelines, and enabling the FAIR reporting of the rapid cycle outputs.

The small-scale prototype includes, in this version:

• Common data models for both the demonstration pilot and PHIRI use cases A to D (including their respective synthetic datasets) (see links in Additional notes)
• Data quality analysis scripts for all use cases
• Data analysis scripts for the demonstration pilot and PHIRI Use Cases A, B, C, and D

In this version of Deliverable 7.2 (14/04/23), we complemented the development of the mid-size prototype of PHIRI Federated Research Infrastructure with the development of a demonstrator piloting the human-to-machine interface to automatically build an updated version of the international comparison report for use case B each time a partner completes the local analysis and shares their aggregated outputs. The containerized application supporting the use case B demonstrator is deployed in IACS servers managed by the Biocomputing Unit (Aragon, Spain). The demonstrator (use case B) can be accessed via links in the Health Information Portal within the menu tab for ‘Federated Demonstrators’ (https://www.healthinformationportal.eu/services/federated-demonstrators) or directly at a) https://phiri.iacs.es/upload for the ‘Upload’ interface -to upload new aggregated outputs from the local analysis produced by a partner participating in use case B, b) at https://phiri.iacs.es/report for the ‘Interactive report’ on international comparative analysis of participants in use case B, and c) at https://phiri.iacs.es/ for an interactive mockup of the web application interface enable upon PHIRI app local deployment in a partners’ systems.

An additional piece of task 7.2, included in the specification of WP7, was to explore the use of the Observational Medical Outcomes Partnership (OMOP) Common Data Model (CDM), a widely adopted CDM for observational health data analytics for representing the information requirements specified in the use cases’ data models to explore the possibility of harmonizing the information requirements of all use cases using a single common data model.

*Rapid-cycle analysis is enabled by participant partners deploying this prototype through continuous data updates.

-------

The current technical stack of the PHIRI FLI includes (x2) Docker containers, using:

• (x1) Server-side (Back-end)
  • Framework: Nest (NestJS) -> TypeScript (https://nestjs.com/)
  • Execution environment: Node.js® >= v14.0.0 (https://nodejs.org/es/)
  • Database management system: SQLite (https://www.sqlite.org/index.html)
  • Data analysis environment: R version 4.0.4 (https://www.r-project.org/)
  • Data wrangling and management: Python (>= Python 3.8.12 (https://www.python.org/))
• (x1) Client-side (Front-end Web application)
  • Web server/proxy: Nginx (https://www.nginx.com/)
  • Framework: Vue.js -> HTML, CSS, and JavaScript/TypeScript (https://vuejs.org/)

General requirements for installing Docker (support)

• Server: 1 server (or VM)
• CPU: 1 CPU minimum (or 2 CPU for VM configuration)
• GPU: No dedicated GPU is required
• RAM Memory: 4 GB minimum (8 GB recommended)
• OS: Unix-based (Linux or other) with Docker support

Specific PHIRI FDI Docker deployment requirements

• RAM Memory: 8 GB minimum (16 GB recommended)

The current technical stack of the Demonstrator includes (x1) Docker container, using:

• (x1) Server-side (Back-end)
  • Framework: Wildfly v.16 (https://www.wildfly.org/)
  • Execution environment: Java, Java Runtime Environment (JVE), Java Persistence API (JPA) (latest versions - https://www.oracle.com/java/)
  • Database management system: PostgreSQL (https://www.postgresql.org/)
  • Data analysis environment: R version 4.1.0 (https://www.r-project.org/)
  • Interactive reports: Quarto version 4.2.0 (https://quarto.org/)
  • Data wrangling and management: Python (>= Python 3.8.12 (https://www.python.org/))
  • Web server/proxy: Nginx (https://www.nginx.com/)

Links to other deliverables: 

Common Data Models for the different use cases (included in the PHIRI app):

• PHIRI - WP7 - Pilot Common Data Model https://doi.org/10.5281/zenodo.5148467
• PHIRI - WP6 - Use Case A Common Data Model https://doi.org/10.5281/zenodo.5148014
• PHIRI - WP6 - Use Case B Common Data Model https://doi.org/10.5281/zenodo.5148022
• PHIRI - WP6 - Use Case C Common Data Model https://doi.org/10.5281/zenodo.5148032
• PHIRI - WP6 - Use Case D Common Data Model https://doi.org/10.5281/zenodo.5148040

RMarkdown scripts for analysis of the different use cases (included in the PHIRI app):

• PHIRI - WP6 - Use Case A scripts for local analyses (R Markdown) https://doi.org/10.5281/zenodo.6377096
• PHIRI - WP6 - Use Case B scripts for local analyses (R Markdown) https://doi.org/10.5281/zenodo.6359893
• PHIRI - WP6 - Use Case C scripts for local analyses (R Markdown) https://doi.org/10.5281/zenodo.6380734
• PHIRI - WP6 - Use Case D scripts for local analyses (R Markdown) https://doi.org/10.5281/zenodo.6377112

Note 1: The PHIRI FRI includes deployment instructions as a PDF document ("deploy_phiri_app.pdf") 

Note 2: The PHIRI FRI includes virtual machine installation and deployment instructions as a PDF document ("PHIRI_WP7_vm_installation_guide.pdf") 

############################################################################################

#### PHIRI app v.2.2.3 changelog #### 

• Fixed RMarkdown analysis scripts related to socio-economic level for use cases A, B and D. 
• Fixed RMarkdown analyses script for use case C to show 0.0 or NA values as empty values.
• Fixed the data syntax checker during the data mapping process that caused problems in required numeric variables when non-numeric values were present without correctly reporting the error 
• Fixed RMarkdown analysis script for use case D; the aggregated function had an unexpected behaviour in periods with null information.
• Fixed RMarkdown analysis script for use case D to identify anxiety and depression for other ontologies such as ICD9 or SNOMED.
• Fixed RMarkdown analysis script for use case D to dynamically consider whether the dataset contains data from 2021 (optional data).
• Update: Do not set the maximum value of the "period" variable in the mapping and checking script to support successive periods.
• Update: Do not set the maximum value of the "age" variable in the mapping and checking script.
• Update: Fixed axes of the graphs (y = 0)
• Update: Minor change in the visualization of figure legend and axes labels 
• Fixed: Removed the filter that took into account the type of treatment to create the selection of patient

#### PHIRI app v.2.2.2 changelog #### 

• Fixed RMarkdown analyses scripts for use cases A, B, and D to handle variables (recommended) with all values missing (empty vars). 
  • Updated the RMarkdown analyses scripts to introduce checkpoints before proceeding with the analyses and continue running even if not all recommended variables are present. 
• Removed the 'Actions' icons in the 'Outputs retrieval' menu interface disabled in certain deployment settings.
• Substituted all 'Actions' icons within the user interface with labelled buttons. 
• Added PHIRI app versioning information in the collapsible menu (left side)

#### PHIRI app v.2.2.1 changelog #### 

• Fixed data syntaxis checker during the data mapping process affecting the expected data model in variables with restricted values when the entire variable is missing 

#### PHIRI app v.2.2.0 changelog #### 

• Fixed data syntaxis checker during the data mapping process affecting the expected data model for the Demonstration Pilot
• Improved display of previous analyses in the Outputs. Now outputs are presented sorted by date of generation. 
• Updated Demonstration Pilot analyses
  • Added supplementary methodological notes to the output HTML report
  • Added generation of aggregated data outputs
  • Refactored R code to facilitate maintenance 

#### PHIRI app v.2.2.4 changelog #### 

• Fixed RMarkdown analysis scripts related to socio-economic level for use case A

#### PHIRI app v.2.2.5 changelog #### 

• Fixed a bug that erroneously caused some data model variables to change their data type. This bug only affected use case A

#### PHIRI app v.2.3.0 changelog ####

• Added use cases A, B, and D data models exploratory mapping to OMOP-CDM at both variable and value levels as a separate ZIP file
• Added information and links to the human-to-machine interface demonstrator (use case B) in the description
• Added information on the current technical stack of the demonstrator (use case B)
• Updated title and description to fit progress in scope