Solution | Pharmacovigilance#

Overview#

Business case#

Post-market drug safety surveillance plays an important role in monitoring the benefit-risk profile of marketed drugs. Effective pharmacovigilance helps organizations identify potential adverse drug reactions (ADRs), prioritize investigations, and support patient safety and regulatory compliance. Failure to recognize and address safety concerns can lead to patient harm, drug recalls, and high costs.

As the volume, variety, and frequency of safety reports grow, drug manufacturers, research organizations, and regulatory bodies need consistent ways to prepare and analyze the data. Automating repeatable tasks gives safety teams more time to investigate potential signals and assess their clinical significance.

This Solution combines automated retrieval of FDA Adverse Event Reporting System (FAERS) data, a reproducible preparation Flow, and interactive dashboards. Users can filter reports, compare cohorts, calculate disproportionality metrics, and explore drug-reaction patterns. Guided Project Setup configures the project, while the dashboards provide an interface for reviewing results and identifying areas for further investigation.

Installation#

  1. From the Design homepage of a Dataiku instance connected to the internet, click + Dataiku Solutions.

  2. Search for and select Pharmacovigilance.

  3. If needed, change the folder into which the Solution will be installed, and click Install.

  4. Follow the modal to either install the technical prerequisites below or request an admin to do it for you.

Note

Alternatively, download the Solution’s .zip project file, and import it to your Dataiku instance as a new project.

Technical requirements#

To use this Solution, you need:

  • A Dataiku 15.2.0+ instance.

  • A configured storage connection. The default setup uses filesystem_managed. Project Setup also provides Snowflake and PostgreSQL connection options.

Data requirements#

The Solution uses quarterly FAERS reports and a supplied reference for FDA product names. Two managed folders store these inputs:

Input Files stores the ASCII files retrieved automatically from the FAERS quarterly data endpoint. Select an available year and quarter in Project Setup; the Flow downloads the source files and processes demographic, drug, reaction, indication, and outcome data. No manual FAERS file upload is required.

Product Drug Names contains the supplied Products.txt reference from Drugs@FDA. The Flow uses this reference to standardize drug names. It’s a separate input and isn’t refreshed by the quarterly FAERS retrieval.

Workflow overview#

You can explore the Solution in the Dataiku gallery.

Pharmacovigilance Flow showing the ingestion, preparation, analysis, and visualization zones.

The project follows these high-level steps:

  1. Configure the project and retrieve data for a selected FAERS quarter.

  2. Clean, standardize, and combine the report data.

  3. Filter reports and create cohort subpopulations.

  4. Calculate disproportionality metrics for drug and adverse event pairs.

  5. Explore reporting patterns and potential signals in the dashboards.

  6. Use the results to guide further safety investigations.

Walkthrough#

The walkthrough covers Project Setup, the preparation and analysis Flow, and the dashboard views used to explore results.

Note

In addition to reading this document, it’s recommended to read the wiki of the project before beginning to get a deeper technical understanding of how this Solution was created and more detailed explanations of Solution-specific vocabulary.

Configure the project and retrieve FAERS data#

From the project homepage, open Project Setup and complete the initial configuration:

  1. Select an available FAERS year and quarter.

  2. Keep filesystem_managed for the default storage setup. If you select Snowflake or PostgreSQL, run Reconfigure Flow Connections before building the Flow.

  3. Choose whether to anonymize manufacturer and drug names.

  4. Run Build the Flow and wait for completion.

Project Setup controls for the FAERS year and quarter, storage connection, and anonymization options.

The API ingestion recipe retrieves the selected quarterly archive from the FDA endpoint over HTTPS. It replaces the previous selection’s files in Input Files, then the Flow parses and prepares the reports automatically.

Project Setup configures the current project; it doesn’t create a new Dataiku Application instance for each analysis. Reuse the configured project for subsequent analyses. If teams need independent data and settings, duplicate or import a separate project and review its setup values.

Prepare the data and generate statistics#

The Flow turns the source files into datasets for analysis through a series of preparation zones.

In Input Files and Data Preparation (Drug), recipes parse the source files, retain the fields needed for analysis, and combine drug information with indications. Visual Split, Group, and Join recipes make these transformations available for review and adaptation.

Input and drug preparation zones in the Pharmacovigilance Flow.

The Data Standardization (Drug) zone combines the prepared drug information with FDA product names. The demographics preparation cleans age, country, and date fields, retains the latest available case version within the selected quarter, and joins report characteristics with reactions and outcomes. Further steps derive event seriousness, apply the selected anonymization options, and consolidate repeated detail rows.

Drug standardization and demographic preparation zones.

Once the base Flow is built, choose report filters such as date range, country, reporter, age, and drug or event exclusions. Run Filter Subpopulation to apply these selections and create the female and male populations used for comparison.

Next, choose the events, drugs, or manufacturers to examine and set the warning thresholds. Run Generate Statistical Metrics to calculate reporting odds ratio (ROR), proportional reporting ratio (PRR), and observed-to-expected ratio (O/E) for the selected drug-event pairs.

Statistical analysis and visualization preparation zones.

The Data Analytics/Statistics and Visualizations zones produce the overall and subgroup results, add configured warning flags, and prepare the datasets used in the dashboard. The project wiki explains the formulas, preparation rules, and interpretation limits in more detail.

Explore the dashboards#

Open Pharmacovigilance analytics after the build and analysis stages finish. Its five tabs combine descriptive summaries with statistical views so users can move from general reporting trends to individual drug-event pairs.

Metrics provides a starting point, with report counts after preprocessing and top values for drugs, adverse events, and manufacturers.

Safety Report Analytics shows reporting patterns by gender, age, seriousness, outcome, and reporter. Use dashboard filters to focus the charts on the characteristics relevant to your analysis.

Safety Report Analytics tab with demographic and report-characteristic charts.

Disproportionality Analysis presents drug-event pairs using ROR, PRR, and O/E. Alongside the native charts, the embedded Drug Explorer lets users select a drug, examine its treemap and sortable metrics table, and download the selected results. Drug selection updates the explorer immediately without running a scenario or changing the surrounding overview charts.

Drug Explorer with a drug selector, treemap, disproportionality metrics, and download button.

DPA by Gender compares reporting patterns and potential signals in the female and male populations. Additional views of age and seriousness help users examine the characteristics of the underlying reports.

Safety Report Signals brings together the preselected drug-event pairs and warning fields based on the configured ROR and PRR thresholds. Use these results to prioritize follow-up review.

Safety Report Signals tab showing drug-event pairs and warning views.

The dashboard supports signal exploration. A reporting association or warning flag doesn’t establish that a drug caused an event or indicate its incidence in the treated population.

Refresh the analysis#

The project scenarios rebuild the relevant outputs when source data or analysis settings change. Run dependent stages in order:

  • New source quarter or anonymization options: run Build the Flow, followed by Filter Subpopulation and Generate Statistical Metrics.

  • Different report or population filters: run Filter Subpopulation, followed by Generate Statistical Metrics.

  • Different event-frequency, drug/manufacturer, or warning-threshold selections: run Generate Statistical Metrics.

Reopen the dashboard after the scenarios complete to review the updated results. Selecting another drug in the Drug Explorer only changes its displayed results; it doesn’t rebuild the underlying data. Filtered reports and selected drug results can be downloaded for further analysis.

Adapt the Solution to your data#

This project gives drug safety and surveillance teams a starting point for preparing spontaneous reporting data, comparing cohorts, and investigating potential ADR signals. Its visible Flow, recipes, and documented analysis steps can be adapted to additional public or private safety sources.

Teams can extend the project to support their own therapeutic areas, reporting requirements, and investigation workflows. Prepared datasets can also feed downstream analyses or case-management systems with the appropriate integration and validation.

By applying this approach across teams, organizations can:

  • Review drug-reaction reporting patterns consistently.

  • Prioritize potential signals for clinical investigation.

  • Compare reporting patterns across populations.

  • Support patient safety monitoring and regulatory compliance activities.

  • Reuse and adapt a common analytical workflow as new data becomes available.

This documentation has provided several suggestions on how to derive value from this Solution. Ultimately however, the “best” approach will depend on your specific needs and data. If you’re interested in adapting this project to the specific goals and needs of your organization, Dataiku offers roll-out and customization services on demand.