FHIR-PYrate

FHIR-PYrate extracts and structures electronic health record data from FHIR servers for clinical research, enabling cohort assembly and retrieval of FHIR resources, imaging studies, and clinical documents for downstream analyses such as deep learning survival modeling.

Key Features:

• FHIR Server Integration: Queries FHIR servers using the FHIR REST API and leverages FHIR search capabilities for uniform resource retrieval.
• Data Structuring: Converts FHIR resources into pandas DataFrames for analysis within Python environments.
• Imaging and Document Retrieval: Downloads imaging studies and filters clinical documents for inclusion in research datasets.
• Parallelization and Filtering: Supports parallelized processing and configurable filtering to accelerate large-scale data handling.
• Customizable Workflows: Adapts workflows to different FHIR servers and available clinical data to match study-specific requirements.

Scientific Applications:

• Cohort Assembly with ICD-10: Identify patient cohorts by extracting ICD-10 codes from FHIR resources.
• Imaging-based Survival Analysis in Breast Cancer with Lung Metastases: Retrieve survival information, thoracic CT scans, TNM staging, and marker positivity to support deep learning–based survival analysis.
• Clinical Data Extraction: Aggregate FHIR resources, clinical documents, and imaging studies for downstream analytic workflows.

Methodology:

Queries FHIR servers via the FHIR REST API; transforms FHIR resources into pandas DataFrames; downloads imaging studies and filters clinical documents; performs parallelized processing and filtering; identifies cohorts using ICD-10 codes; and applies a deep learning model using CT scans, TNM staging, and marker positivity for survival analysis.