Understand The Clinical Narrative
By Ken Congdon, editor in chief, Health IT Outcomes
A few weeks ago, I wrote an article titled Save The Clinical Narrative that emphasized the importance of preserving the medical narrative to aid in clinical decision making. I received a great deal of feedback on this article from physicians, healthcare IT personnel, and the vendor community alike. Perhaps the most interesting feedback I received came from Michael Finke, CEO of M*Modal, a provider of speech understanding products and services. While Mr. Finke agreed with the premise of my article, he felt my explanation of the value of the clinical narrative was incomplete.
"While the clinical narrative is an important way to describe the patient's story, saving the narrative simply for the narrative's sake offers limited value," says Finke. "We need to save the narrative for the sake of quality documentation, and that can only be done if you understand the narrative in a significant manner."
Understanding The Clinical Narrative Requires More Than Natural Language Processing
So what does it mean to understand the clinical narrative, and what tools are necessary to gain this insight? According to Finke, a complete patient record requires both structured data (e.g. field data captured by an EMR) and unstructured data (e.g. dictated patient notes that make up the clinical narrative). However, if this data is kept separate and simply isolated in silos within the same record, its significance is weakened. This is the problem with traditional natural language processing technologies according to Finke. These tools simply tag the narrative as part of the record. To realize the full power of the narrative, Finke suggest that healthcare facilities need to make the data contained in the narrative accessible to the EMR and tap into that information as part of their clinical support, coding, and compliance processes.
"Accessing clinical narrative data effectively in today's world requires more than just tagging using natural language processing technology," says Finke. "Instead, users should be empowered to query their documentation for answers to clinical questions. In other words, I should be able to query all narrative documentation to identify all patients with a particular heart defect or other medical symptom by entering a few simple questions. This functionality really puts the information contained in the narrative to work for the physicians."
This type of query-based documentation environment is exactly what M*Modal has apparently designed with its AnyModal Conversational Documentation Service (CDS). This service is a SaaS (software-as-a-service) platform that not only encodes all clinical documentation, but makes them a seamless part of your existing clinical workflows. Furthermore, the system is designed to access the data contained within the narrative documentation to answer clinical questions. It does this through intelligent processing technology that "learns" from every interaction.
"It's not like Google," says Finke. "It's not trying to find content based on keywords. Instead, its leverages clinical ontologies and natural language processing algorithms to answer meaningful questions against structured and unstructured content."
While this level of sophistication may have been overkill in the past, Finke believes this type of intelligent documentation platform will be increasingly important within the next few years due to the exponential increase of electronic data that physicians will be expected to manage in light of EMR adoption. "Physicians will need an easier way to weed through all of this data to isolate the information they need to make the most informed decisions. This will be accomplished through a combination of natural language processing, semantics, and inference technologies. Think of it as knowledge management for the physician of the ARRA era."
Ken Congdon is Editor In Chief of Health IT Outcomes. He can be reached at ken.congdon@jamesonpublishing.com.