Pathling

Pathling provides scalable analytics for FHIR clinical data by integrating FHIR resources with clinical terminologies such as SNOMED CT to enable exploratory data analysis, patient cohort selection, and data preparation.

Key Features:

• Apache Spark foundation: Built on Apache Spark to enable scalable distributed processing of large healthcare datasets.
• Server implementation with FHIR API: Implements a server-side FHIR API that exposes novel operations specifically for exploratory data analysis (EDA), advanced patient cohort selection, and comprehensive data preparation.
• FHIR Terminology Service integration: Integrates with a FHIR Terminology Service to resolve and use clinical terminologies within queries.
• Support for SNOMED CT: Enables incorporation of SNOMED CT clinical concepts into analytics workflows and queries.
• FHIR standards integration: Supports analysis of FHIR resources alongside complex clinical terminologies within analytical workflows.

Scientific Applications:

• Exploratory Data Analysis (EDA): Perform EDA directly on FHIR resources using specialized API operations.
• Patient cohort selection: Define and extract advanced patient cohorts from FHIR data using terminology-aware operations.
• Data preparation: Prepare and transform FHIR data for downstream analytics and analytics-enabled applications.
• Terminology-aware analytics: Incorporate detailed clinical knowledge from terminologies like SNOMED CT into analytical queries and results.

Methodology:

Pathling is implemented on Apache Spark, exposes a server-side FHIR API with operations for EDA, cohort selection, and data preparation, and integrates with a FHIR Terminology Service to use terminologies such as SNOMED CT.