Tools and integrations
Working with HARK from common research environments.
The resource is designed to be reachable from the environments audiology and informatics researchers already use: Python and R clients, a documented SQL surface, REST endpoints, and notebook workflows on BDSP. Listed integrations are at varying stages of development; capabilities and stability are noted on each card.
Jupyter Notebook Templates
Ready-to-run notebooks for cohort discovery, normative-curve queries, and audiogram trajectory analysis against a local HARK sample.
HARK Agent (LLM)
Natural-language interface that translates clinical questions into HARK queries and returns plots, tables, and citations.
pyhark Python SDK
Typed Python client for the HARK REST API with pandas-native returns and built-in de-identification guards.
POD-Vis
One of several interfaces under development. POD-Vis (Probing Outcomes Data with Visual Analytics) provides a graphical view of normative curves and asymmetry references for users who do not query the data programmatically.
harkr R Package
Tidyverse-friendly R bindings for HARK, designed for epidemiologists and biostatisticians working in RStudio or Posit Cloud.
BDSP-hosted compute
Approved investigators run analyses inside the Brain Data Science Platform against the same release the consortium uses, with Python tooling, optional GPU access, and tiered controlled access.
hark CLI + REST API
Command-line tool and OpenAPI-documented REST endpoint for scripting, CI pipelines, and lightweight integrations.
Open standards, additional tools welcome.
HARK is intended to publish its schema, ontology crosswalks, and REST API as open standards, so that tools speaking FHIR, SQL, or Python can read the resource. The consortium plans a community contribution track for additional connectors (for example, Stata, SAS, Observable, Streamlit, or MATLAB).
7 first-party integrationsopen standardsApache 2.0 + CC-BY 4.0 license