Using Google Cloud and Augmedix technologies, HCA Healthcare is exploring potential ways for integrating generative AI to automate administrative tasks so that clinicians can spend their time caring for and interacting with patients.

The healthcare provider is one of the largest in the U.S. with 182 hospitals and approximately 2,300 ambulatory care sites in 20 states and the U.K. It is seeking to apply generative AI to automate medical documentation, patient handoffs among nurses, and caregiver support. Here are three ways it is doing this:

• Documenting conversations: As part of a pilot program at approximately 75 ERs at four HCA Healthcare hospitals, doctors are writing down key medical information from conversations with patients using Google’s AI technology and the Augmedix app, designed for ambient medical documentation.
  
  The app is on a hands-free device that allows users to record data that is instantly converted into medical notes with the app’s natural language processing and Google Cloud’s generative AI’s multiparty, medical speech-to-text processing. The physician then reviews and transfers the notes in real time to the hospital EHR.
  
  Doctors within the pilot have shown overall satisfaction. All three companies are refining this solution, and HCA Healthcare will implement it in more hospitals later this year.



• Automating handoff reports: Compiling reports for other nurses on patients is time-consuming and manual and varies in detail. But using one of Google Cloud’s large language models (LLMs), nurses are able to automatically produce these summaries, saving themselves time while maintaining continuous patient oversight.
  
  Prompts direct the LLM to prioritize details, such as medication changes, lab results, vital sign fluctuations, patient concerns, and overall treatment response. The model’s outputs are intuitive so nurses can easily read and act upon the data. UCF Lake Nona Hospital in Florida is testing the prototype, and nurses have shown a desire to use it in practice based on the speed, accuracy, and relevance of its draft reports.



• Answering complex medical questions for caregivers: Google Cloud’s Med-PaLM 2 LLM uses medical domain data to answer complex medical questions and issues long-form answers to consumer health questions, which may be helpful in certain critical use cases. It has an accuracy of 86.5% on U.S. Medical License Exam-style questions.
  
  “We expect Med-PaLM 2 will be especially useful when we’re asking complex medical questions that are grounded on scientific and medical knowledge while looking for insights in complicated and unstructured medical texts,” said Dr. Michael Schlosser, of care transformation and innovation at HCA Healthcare, in a statement.

The tests are an expansion on HCA Healthcare's existing partnership with Google Cloud, formed in 2021.

Health IT Homepage