INquery: Phoenix Children's

Phoenix Children’s has partnered with the Arizona State University Cloud Innovation Center (CIC) to build a tool that lets administrators query medical documents using natural language processing.

For a health system of this size, it is a challenge to categorize and query the many regulatory policy documents that staff must be aware of. Medical facilities allocate manual resources to sort through cluttered document systems which results in long lead times for administrative tasks. When trying to ensure that patients receive the best possible treatment, an organized and easily accessible search system for medical documents can be invaluable. 

For example, insurance company guidelines on procedure pre-authorization are constantly changing. To prevent patient families from unexpected charges or denial of claims, Phoenix Children’s administrators manually validate insurance websites for pre-authorization requirements for every procedure. 

The “INquery” tool will allow Phoenix Children’s administrators to upload large medical policy documents (such as insurance pre-authorization guidelines) to an Amazon S3 bucket as a central repository. Amazon Sagemaker Foundation AI generative model will be leveraged to generate embeddings while storing and querying the PDF documents. This will allow Phoenix Children’s to efficiently use natural language while searching procedure specific insurance company pre-authorization guidelines, thereby reducing the time and resources required to verify information prior to medical procedures.