LoginSign Up
← Back to all sparks
N

npi

ANALYTICS
Velocity0.0

Access the U.S. National Provider Identifier (NPI) Registry Public Search API

github.com/ropensci/npi ↗

An R client for the US provider registry, tightening its types and edge-case handling

healthcare-datar-packageapi-clientdata-validation
◆Current state
npi wraps the US National Provider Identifier registry API for R users, covering search, validation and summarisation of provider records. The one release in view is a consolidation pass rather than new surface: input normalisation, vectorised validation, and a typed empty result instead of an ambiguous one when a search finds nothing.
◆Where it's heading
The work is aimed at making the package behave predictably inside larger pipelines. Returning a typed empty `npi_results` object on no matches, and having `npi_is_valid()` accept vectors and return logical vectors, both remove branches a caller would otherwise write by hand. The bug fix follows the same line — `npi_summarize()` no longer drops input rows when a record's nested address or taxonomy data is missing.
◆Prediction
With a single release visible there is not enough of a pattern to predict a direction confidently; the changes here suggest continued interface tidying rather than new API coverage, but that is one data point.

◆Recent moves

  1. 1mo ago

    Vectorised validation and typed empty search results

    Interface hardening across the board: `npi_search()` normalises `address_purpose` to API constants and accepts case-insensitive input, `npi_is_valid()` becomes vectorised, and an empty search returns a typed `npi_results` object rather than something callers must special-case. A fix stops `npi_summarize()` dropping rows when nested address or taxonomy data is missing.

    View source ↗