Semantic interoperability and price analytics in hospital transparency data: a multi-stage pipeline with NLP and machine learning.

PURPOSE The U.S. Hospital Price Transparency mandate requires public disclosure of machine-readable files (MRFs), yet profound data heterogeneity hinders their utility for research and consumer use. This study evaluates a novel, multi-stage computational pipeline to systematically process diverse MRFs and enable robust price analysis.

METHODS The pipeline integrates a configurable parsing engine with an NLP module using Sentence-BERT embeddings and K-Means clustering for semantic standardization of procedure descriptions and CPT code alignment. It was applied to MRFs from five U.S. hospitals (Mayo Clinic, Johns Hopkins, Stanford, Jackson Memorial, and Mass General) for five elective procedures.

RESULTS The pipeline successfully processed 7,449 records, revealing substantial price variation across hospitals for semantically equivalent services. Predictive modeling using Lasso regression yielded an R^2 of 0.70 (RMSE = $808; MAE = $533).

CONCLUSIONS This work advances scalable, semi-automated MRF research methodology, transforming opaque pricing data into an analytically tractable form with implications for healthcare policy and consumer tools.