Artificial Intelligence, Health IT, SYN

Google rolls out AI tools to review unstructured medical text

Google Cloud developed a suite of tools using natural language processing to help healthcare workers review and analyze medical documents more easily.

Google’s tool also lets clinicians develop their own models to extract information from clinical records. Photo credit: Google

Google unveiled two AI tools to help clinicians review and analyze medical documents. In a blog post on Tuesday, the company shared it has developed a Healthcare Natural Language API to help identify important insights, such as vaccination or medication information, from unstructured text. Google Cloud also developed a tool called AutoML Entity Extraction for Healthcare to make it easier for healthcare professionals to build their own tools to extract clinical information from health records.

The goal is to make it easier to pull in important information that might not be in the structured fields of a medical record, but are instead buried in free text notes. For example, medical refill requests from a call center conversation, or information on substance abuse or physical activity are often found in unstructured text.

It works by using natural language processing to identify clinically relevant attributes based on the surrounding context, Andreea Bodnari, a product manager for Google Cloud, wrote in a blog post. It’s able to distinguish previous prescriptions from medications prescribed for the future, and can distinguish between information about a patient from information about a patient’s relative, she added.

Google unveiled two tools using natural language processing to help clinicians review unstructured text from medical records. Photo credit: Google

The technology could also prove useful in other contexts, such as helping better match patients again inclusion or exclusion protocol criteria for clinical trials, or pulling relevant symptoms and medications from a transcribed telehealth visit.

In an emailed statement, Google said it had customers, “such as large hospital centers,” who are using the technology.

It’s one of many efforts by Google Cloud to develop back-end tools to support clinical work. Earlier this year, Google invested $100 million into telehealth company Amwell in a private placement. As part of the deal, Google struck a strategic technology collaboration with Amwell, floating potential ideas such as using AI for patient intake or collecting insurance information.

A separate partnership with Ascension Health stirred more controversy. Google was working with the health system to develop a search tool for health records, so that clinicians could search for an MRI or find a vaccine record.

Shares1
Shares1