Walking out of your patients room, you pull out your smartphone to dictate a quick progress note. You note that your patient, hospitalized for a hip fracture, is complaining of bilateral leg swelling and orthopnea.
Once you complete your dictation, you are immediately presented with the transcribed text.
In addition, you see two flags. The first notes that the patient is not on DVT prophylaxis and suggests that given the patients age, weight, and co-morbidities, you use mechanical prophylaxis while also displaying appropriate references for that recommendation.
The second flag suggests that you consider a stress cardiomyopathy, noting that there are numerous case reports and series of patients who match your patients demographics developing this condition after significant trauma.
When it comes to the way we work, medicine is definitely a late adopter as far as technology goes. One need not look further than the stacks of paper charts at hospital nursing stations or the alphanumeric pagers at the waist of most physicians. A new partnership between Nuance Communications and the University of Pittsburg Medical Center (UPMC) seeks to help our profession not only catch up, but take the lead in applying cutting edge technology to real-world problems, making the above scenario a realistic possibility.
We had the opportunity to talk to both Dr. Rasu Shrestha MD/MBA, VP of Medical Information Technology at UPMC, and Joe Petro, Senior Vice President of Healthcare Research and Development at Nuance, about this endeavor.
Nuance is certainly well known for its speech recognition tools, enabling highly accurate and real-time dictation via its Dragon software suite. The power of this technology, however, goes well beyond just documentation, particularly with its medical intelligence software with which
“…clinical data is transformed into actionable information that healthcare providers and payors can use to optimize the value of electronic health records (EHRs), drive better patient care, improve quality reporting and compliance, and enhance financial performance.”
According to Dr. Shrestha, “[UPMC] has upwards of 39 different systems that we have been able to interoperate…80% of that data is unstructured.” Much of this data is housed in a single place, the MARS repository, which stores nearly three decades of data. As Dr. Shrestha points out, “It’s not just about the output, its also about the input – the data you put in to teach it.”
The power of this data is certainly not lost on UPMC, as Dr. Shrestha notes that they began looking for a way to use that data several years ago. With Nuance, according to Dr. Shrestha, UPMC found the complete package of front-end and back-end software to collect, index, and process the mountains of data collected by this 50,000-strong healthcare behemoth.
Through this Joint Development Agreement, UPMC will essentially be both a partner and test bed for development of three use cases of Nuance’s software:
- Tools for real-time clinical documentation improvement
- Tools for converting unstructured documentation into structured, discrete data
- Tools for populating structured data into EHRs (focused on meaningful use)
According to Mr. Petro, this partnership will “make it easier…to take any kind of free text, process it through the cloud, and what comes back is an .xml-type structured document that contains all of the facts from the initial document.” He does note though that is this just the beginning, “the first inch of a massive iceberg.”
The earliest products that will be commercially available will focus on clinical documentation and quality analytics with a focus on meaningful use. The former will be using 3M’s platform for generating document-specific queries for more precise documentation, particularly important with the increasing emphasis on metrics like severity of illness and payment models like pay for performance. The latter will enable populating EHRs with discrete data that can be analyzed and used to meet meaningful use criteria.
The future applications of this technology are far more expansive. The ability to create discrete data from unstructured documentation can help create enormous datasets that can provide new physicians with new insights into disease, assist in creating more accurate prediction rules, and more.
This data, in combination with resources like best practice guidelines, clinical trials, case reports, and expert consensus recommendations can help power tools that provide real-time decision support at the point of care. And it can do so both at the right time when clinical decisions are being made and with minimal disruption of the clinicians workflow.
Add into this mix the analytic and processing power of IBM’s Watson, and the potential transformative power of this technology is practically boundless. With the partnerships between Nuance and IBM, UPMC, Cedars-Sinai Comprehensive Cancer Center, Mr. Petro describes the deliberate approach he and his colleagues are taking to get there. “Identifying the right problem is key,” Mr Petro says, noting that Nuance is identifying specific use-cases where the context in which questions are posed can be tightened, say to an oncology population or single institution.
One of the challenges Mr. Petro highlights is the silos of data created by incompatible EHRs, radiology systems, and other clinical IT systems. He points out though that overcoming this barrier may ultimately require legislation or regulation.
The underlying technology – clinical language understanding and medical intelligence from Nuance and question answering from IBM – is powerful, particularly in the context of the incredible amount of information available in medicine. As Dr. Shrestha puts it, “Marry that with real-time clinical scenarios…that creates a powerful tool to improve outcomes.”