While clinicians have often expressed frustration over the way they have to interact with EHRs, Google is working on technology for streamlining functions like data searches and predictive text search, according to Google artificial intelligence head Jeff Dean, PhD.
During a recent episode of a podcast by Eric Topol, MD, and Abraham Verghese, MD, "Medicine and the Machine," Dr. Dean discussed his predictions for how EHRs will evolve in healthcare and some of Google's current projects.
Here are six insights from Dr. Dean, cited in an Aug. 20 Medscape report.
1. Google has worked with other organizations on using deidentified data to refine EHR searches in a way similar to how the tech company trains natural language models, Dr. Dean said. With the natural language models, the researchers aim to use the prefix of a piece of text to predict the next word or sequence of words that is going to occur.
2. An example of natural language models would be a model applied to email messages, so when a person is typing out a message, the AI suggests how they might complete the sentence to save typing, Dr. Dean said.
3. Google is working with the same approach to give clinicians suggestions about what might occur next in the EHR for a particular patient, Dr. Dean said, adding, "If you think about the medical record as a whole sequence of events, and if you have de-identified medical records, you can take a prefix of a medical record and try to predict either the individual events or maybe some high-level attributes about subsequent events, like, 'Will this patient develop diabetes within the next 12 months?'"
4. While the idea of creating an AI model that uses every past medical decision to help inform all future medical decisions is complicated, Dr. Dean said the feat is a "good north star" for potential health IT innovations.
5. Dr. Dean said his group has done some work using an audio recording of a patient-physician conversation to develop a medical note that a clinician can just then edit a little bit instead of having to type up the entire note.
6. Creating summarized notes from conversations might also be a good assistant tool that not only helps reduce clinician burden but could lead to higher-quality data in the EHR, according to Dr. Dean.
"We all know that often clinicians copy and paste the most recent note and don't really edit it appropriately. That's partly because it's very cumbersome and unwieldy to interact with some of these systems, and speech and voice are a more natural way of creating notes," Dr. Dean said.