Natural language understanding software vendor MModal has introduced a new line of products called Catalyst to help clinicians extract unstructured data and combine it with structured information.
Unstructured data consists of dictated notes from patient exams, while structured data includes information on medication, lab values and vital signs, Mike Raymer, MModal’s senior vice president of solutions management, told eWEEK.
Announced on June 12, MModal’s new platform includes two components: Catalyst for Quality and Catalyst for Radiology. Catalyst for Quality allows health care executives to improve the quality of documentation and satisfy the requirements for meaningful use of electronic health records (EHRs), according to the company. Catalyst for Radiology enables radiology professionals to spot important clinical information within radiology documents, comply with regulations and improve quality of care.
Catalyst combines speech recognition and natural language processing while providing tools for annotating and sharing data.
“What natural language understanding does is it’s really a continuous learning technology platform that lets you look at phrases of data structured [according to] how humans interact with one another and apply human life learning to that experience,” said Raymer.
On May 9, MModal introduced Fluency Cloud, a group of speech-to-text applications that populates electronic health records with data from doctors’ narrated accounts of patient interactions.
The products in Fluency and Catalyst are built using data on the company’s Speech Understanding cloud platform, which allows health care organizations to scale their use of the applications as well as access data from anywhere, MModal reported.
The challenge for physicians is gaining the ability to search through unstructured data, and several companies are developing products to help providers make sense of this hidden information in order to make diagnoses. Nuance Communications incorporates Clinical Language Understanding (CLU) applications in IBM’s Watson supercomputer.
Catalyst conforms to vocabularies that allow clinicians to process data and tag specific concepts, said Raymer.
According to the federal government’s guidelines on meaningful use of EHRs, physicians must encode patient data using the Systematized Nomenclature of Medicine-Clinical Terms (SNOMED-CT) medical vocabulary, said Raymer.
In addition, Catalyst supports Logical Observation Identifiers Names and Codes (LOINC), a standard used by doctors to document laboratory and clinical observations, and RxNorm, which provides normalized names for clinical drugs.
By sorting through data stored in the Speech Understanding cloud platform, Catalyst is able to avoid repeating documentation mistakes and can sharpen its coding of terms from SNOMED-CT and LOINC, said Raymer.
Not being able to recover the meaning in doctors’ notes hinders reimbursement, according to Raymer. Identifying the cleaning of a wound as “tissue removal” rather than “debridement” could mean higher reimbursement, he noted.
Catalyst allows doctors to retrieve information on key clinical concepts from documentation. Searchable scenarios may include a “collapsed lung due to medical procedure complications,” MModal reported.
MModal’s NLU engine examines the context of words or events to see if they’ve occurred or not. If they haven’t occurred, they’re considered negated, and if they have occurred, they’re considered active, Raymer explained.
In addition, Catalyst is able to provide context to data since doctors don’t speak the words in the same order, according to MModal.
MModal plans to add additional products to its Catalyst line in the coming months.