Natural Language Processing Technology Will be Critical to ICD-10 Success

It seems October is becoming a critical month for change in the health system. The major provisions of the Affordable Care Act will come to life in October 2013. Twelve very short months later, our transition to the ICD-10-CM/PCS system for coding medical documentation will be complete.

Regarding ICD-10, Optum kicked off this series in April to provide practical information health care organizations can use to make their journey to the new coding standard successful.

This month, we’ll discuss the technology behind computer-assisted coding (CAC), which is one of the key purchases hospitals are making to support their ICD-10 coding efforts. Getting familiar with what’s under the hood of CAC tools is essential to achieving optimal performance in coding applications and beyond.Mark Morsch

Natural language processing (NLP) technology powers CAC applications. The NLP engine is responsible for automatically reading clinical documentation to identify diagnoses and procedures and then recommending the codes assigned to clinical cases.

Five distinct methodologies drive the ability of NLP engines to organize and extract meaning from clinical documentation. In simple terms:

  • Medical Dictionary Matching maps words in the clinical documentation to medical terminology
  • Pattern Matching finds word patterns that describe a diagnosis or procedure
  • Statistical uses pre-coded documents to train and evolve algorithms
  • Symbolic Rules identify codes from language using linguistic rules and symbols
  • Symbolic Rules & Statistical Components – a more advanced approach – uses a hybrid of mathematical modeling and linguistic rules to identify meaning and context

Each approach affects performance of the NLP, and ultimately the CAC applications or other applications it powers. Understanding this variation is essential to maximizing return on investment now, through realizing immediate measureable gains in current coding processes, and is critical to ensuring scalability to broader applications, including ICD-10 clinical documentation improvement (CDI) programs and health analytics.

ICD-10, with its expansive 155,000+ potential codes, puts unique demands on people and machines tasked with indexing diagnoses and procedures to its coding structure. Those unable to decipher and adequately describe granular details of patient care, from laterality to severity and acuity, put accurate payments at risk. In other words, insufficient detail will lead to reduced revenues for physicians and hospitals.

Of the five NLP methodologies in use today, the hybrid Symbolic Rules & Statistical Components approach has proven the best suited to the rigors of ICD-10. The methodology paired with patented “mere-parsing” capabilities enables sophisticated identification and interpretation of words and phrases in their medical context.  This bridges the natural way physicians communicate about patient care to the data required in hospital information systems to help provide consistent and accurate diagnoses and procedure codes for proper reimbursement.

For example, a patient record may include the phrase “breast cancer.” NLP dependent upon simple matching of medical terms and word patterns or pre-programmed statistical analysis can’t determine whether this is a current or past diagnosis, or whether it pertains to the patient or the patient’s family history.

Optum today published Not all NLP is Created Equal: CAC Technology Underpinnings that Drive Accuracy, Experience and Overall Revenue Performance, a white paper describing NLP methodologies and performance variances in more detail. This and a webinar we will host later this month are excellent resources for learning more about NLP technology and gaining confidence in the technology’s ability to support ICD-10.

NLP technology has a vital role to play in addressing ICD-10 and other current issues facing health care organizations. It will also be vital to supporting future needs, from provider-to-provider collaboration to outcomes based reimbursement, facilitated by growing use of electronic medical records, emphasizing the importance of choosing the right technology today.

Resources:

–Mark Morsch, MS, Vice President of Technology at Optum

Leave a Comment

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s