Palantir Technologies (NYSE:PLTR) today issued additional details about the Palantir Foundry capabilities it will be highlighting in its Life Sciences demo at “Double Click” on Wednesday, April 14, 2021, at 11:00am ET.
Palantir’s customers in Life Sciences span pharmaceuticals, research institutions, healthcare systems, and others. These customers rely on Foundry’s secure and collaborative architecture to discover and develop drugs, manage commercial launches, conduct clinical research, and distribute vaccines. For example:
- More than 1,000 researchers at the NIH’s National COVID Cohort Collaborative (N3C) use Foundry to securely maintain one of the largest patient-level data assets of COVID-19 Electronic Health Record (EHR) data in the world and collaborate on more than 150 research projects (as illustrated in this video demonstration by N3C).
- Thousands of General Practitioners at NHS England order, allocate, track, and deliver every one of the 29+ million vaccines administered in the country to date (as outlined in the UK Government’s website for data and insights on COVID-19).
At the first installment of Palantir’s Double Click series, engineers will demo the latest Foundry capabilities, as well as several new Archetypes available on top of Foundry that make this work possible in a matter of clicks, including:
- Reproducible research in any environment. Foundry’s interoperable, modular, and open architecture enables users of varying technical ability to integrate any data, use any tool, manage any model, and write code in any language — while maintaining end-to-end reproducibility, transparency, and security of their analyses. Foundry integrates its governance and knowledge management capabilities with common data science environments such as RStudio or HPC clusters, enabling users to work reproducibly with any libraries of their choice — from specialized bioinformatics tools such as plink to machine learning libraries such as scikit-learn, tensorflow, or MLlib.
- Purpose-based access controls. Foundry’s Purpose-Based Access Controls Archetype enables automated or managed assignment of permissions based on the proposed use of a particular data source or dataset. By systematically encoding data access requests alongside the context of the associated research, these access controls provide governance administrators with granular insight into every analysis a piece of data is used for, by whom, and for what reason.
- Systematic model management. Foundry’s Model Management Archetype supports the governance and deployment of predictive models, including those developed outside Foundry. Researchers can readily reproduce results, track how each version of a model was developed, automatically update models when upstream source data changes, compare performance metrics, and tag versions for release into staging or production environments. Models can be deployed within Foundry (batch or streaming, or API endpoint for ad-hoc querying) or to external systems, including specialized hardware. Health checks, telemetry, and rich model monitoring of deployments help organizations go beyond model drift to understand and manage their models against real-world scenarios.
- Out of the box pipeline templates for standardized data models. Foundry’s Pipeline Archetype deploys pre-built data pipelines that convert common real-world data sources such as Optum claims or TriNetX into standard data models, such as OMOP. This enables researchers to easily pivot between sources, and reuse cohort code lists or analytic methods that have already been defined by the open-source community. For unique data sources, such as historic Randomized Controlled Trials, Foundry’s machine learning-powered Entity Resolution Archetype provides the framework for automatically suggesting mappings to a user-defined data model and enabling human-driven confirmation.
