Rawpixel - Fotolia
Health information management tech plays an important role in population health. In this Q&A, John Showalter, chief health information officer at University of Mississippi Medical Center located in Jackson, Miss., talks about the health information tech used today to further population health, the technology his organization is using and the emerging technologies at play.
How can existing and emerging health information management tech be used to improve advanced analytics, identify target populations, and enhance clinical decisions to support new payment models and population health?
John Showalter: I own three of the main platforms for CDI and for coding HIM type work or computer assisted coding which uses natural language processing, voice to text to replace dictation as well as physician aided documentation or real-time natural language processing as well as some predictive analytics around length of stay and what's the most probable diagnosis, [diagnosis-related group] DRG annotation inside the documents. And we've been using them for a very long time, at least from a technology perspective for healthcare, but they've been used for years. And they've been used to help make coders faster, make administrative processes more accurate. But what we really do is read the electronic health record and read the 80-plus percent of the electronic health record that is not discreet and turn it into actionable information. So if you want to use that to identify your diabetics, you can use it to identify your diabetics. If you want to use it to identify people that have incidental findings … and make sure that they get the care they need, you can do that as well. So these tools that we've been using to create claims have a very real purpose in being used to drive population health.
What is the emerging health information management tech at play here?
Showalter: The emerging technologies are deep machine learning and multi-layered predictive analytics, which combines with the computer assisted coding and the detecting of conditions inside the chart to free up all of that information to be used in predictive analytics. So today, we're using predictive analytics and we're only using it on 5% to 15% of the electronic medical record. By combining it with natural language processing, we can begin to add more data to predictive analytics and thereby combining the predictive analytics with real-time natural language processing with computer systems physician documentation, we can actually get context based alerts back to the physicians so that you can be dictating about how you have restrained a patient and the predictive analytics can note that they're high risk for a pressure ulcer and then pop up to the screen and say, "Hey physician, you just documented this patient is restrained. They're at high risk for pressure ulcers. Would you like to do something about that?"
Are any hospitals or health systems already doing this?
Showalter: We are beginning to work with our predictive analytics vendor and our natural language processing vendor. So we work with Jvion and M*Modal and we are planning to roll out later this fall a detection of exactly that example: Real-time national language processing as the physician documents that the patient has been restrained and alerting them about the pressure ulcer. So we are planning to combine those two in real life pilots later this fall … I think this is really just the beginning. I haven't heard of anyone else doing this.
How could these technologies be used to facilitate population health at a state-wide level?
Showalter: We are not doing this yet but we are in early discussions with our statewide health information exchange that is gathering about 80% of all the radiology records across the state. [W]e're planning to work with our vendors … to natural language process all the radiology reports across the state in identifying people that have a lung nodule that needs to be followed up. [T]hen … someone from the state Department of Health [would] give those patients a call to make sure that they are getting followed up with their cancer screening.
The link between population health and natural language processing
Interoperability essential to achieving population health strategies
Population health best practices for value-based reimbursement
Dig Deeper on Population health management
Related Q&A from Kristen Lee
Two experts agree: AI will become essential to healthcare. AI can help providers step away from the EHR and enable precision medicine. Continue Reading
Vendors demonstrate that interoperability is possible when it comes to patient medication information. This makes CIOs hopeful for the future. Continue Reading
Healthcare should be getting ready for the cloud. At least, that's what one health IT expert thinks. Read his thoughts on cloud computing healthcare ... Continue Reading