Data Scientists on Real-Time Monitoring Platform: “We Have a Direct Impact”
This year at the HIMSS conference, artificial intelligence (AI) and how it can be utilized in healthcare settings, is once again a hot topic. At a panel session entitled Real-Time Analytics Monitoring Platform: Usable AI in Action, three data scientists presented a tool named RAMP, which was created for physicians at UVA Health.
Valentina Baljack, Ph.D., senior data scientist, informed the audience that what sets their medical health system apart is the data science team that includes a variety of backgrounds and disciplines. “When we started thinking about using AI in healthcare, instead of focusing on just one form of it, generative AI, we started thinking about what AI actually is. By its definition, it is the capacity of machine human behavior, such as recognizing patterns or solving problems.”
Clinicians and physicians are familiar with early warning scores, Baljack said. Early warning scores (EWS) are tools that use vital signs to identify patients who may be deteriorating. The data scientists’ team game up with a model that captures about 9 percent of deteriorations early on. They wanted to get clinicians to adopt these models. With the engagement of physicians, the team was able to create reliable predictions.
Baljack explained that data is being collected from different sources. “We are able to deliver our results within five to ten minutes of the change in a patient’s chart. “We have a direct impact on patient care.”
“Apart from our integration with our EHR (Electronic Health System), we have open-source technology,” Margot Bjoring, Ph.D., added. “The majority of our data is coming directly out of the EHR via APIs that they provide to access various patients.” “Being outside of the EHR also gives us the flexibility to add additional external data sources.”
“Probably the most technically challenging part of this project was getting these data sources set up and integrated, especially at the beginning, getting our connection to our EHR setup,” Bjoring explained. “For our provider teams, we have a list of their patients up to date with the current model scores for each of those patients on the team…It's also useful for workload management.”
“The biggest thing that I want to point out is that this is not just a niche application that we developed for one department or one unit in the hospital,” Michael van den Bossche, Ph.D., noted. “This is used throughout our hospital….We use it at acute care units in ICUs. We use it for our neonatal population, for pediatrics, and for the adult population. And although right now we're mostly applying this to medical specialties, we're also in the process of expanding this to surgical specialties as well.”
With early warnings and alerts, van den Bossche highlighted, the goal is to intervene as early as we can.
