Implicit Entity Recognition in Clinical Documents

Proceedings of the Fourth Joint Conference on Lexical and Computational Semantics (*SEM 2015), pages 228–238, Denver, Colorado, June 4–5, 2015. Implicit Entity Recognition in Clinical Documents Sujan Perera*, Pablo Mendes†, Amit Sheth*, Krishnaprasad Thirunarayan*, Adarsh Alex*, Christopher Heid¶, Greg Mott¶ *Kno.e.sis Center, Wright State University, Dayton, OH, USA †IBM Research, San Jose, CA, USA ¶Boonshoft School of Medicine, Wright State University, Dayton, OH, USA sujan@knoesis.org, pnmendes@us.ibm.com, amit@knoesis.org, tkprasad@knoesis.org adarsh@knoesis.org, caheid2@gmail.com, mott11@wright.edu Abstract With the increasing automation of health care information processing, it has become crucial to extract meaningful information from textual notes in electronic medical records. One of the key challenges is to extract and normalize entity mentions. State-of-the-art approaches have focused on the recognition of entities that are explicitly mentioned in a sentence. How- ever, clinical documents often contain phrases that indicate the entities but do not contain their names. We term those implicit entity mentions and introduce the problem of im- plicit entity recognition (IER) in clinical doc- uments. We propose a solution to IER that leverages entity definitions from a knowledge base to create entity models, projects sen- tences to the entity models and identifies im- plicit entity mentions by evaluating semantic similarity between sentences and entity mod- els. The evaluation with 857 sentences se- lected for 8 different entities shows that our al- gorithm outperforms the most closely related unsupervised solution. The similarity value calculated by our algorithm proved to be an effective feature in a supervised learning set- ting, helping it to improve over the baselines, and achieving F1 scores of .81 and .73 for dif- ferent classes of implicit mentions. Our gold standard annotations are made available to en- courage further research in the area of IER. 1 Introduction Consider the following sentence, extracted from a clinical document: “Patient has shortness of breath with reaccumulation of fluid in extremities.” It states that the patient has ‘shortness of breath’ and ‘edema’. The former is explicitly mentioned, while the latter is implied by the semantics of the phrase ‘reaccumulation of fluid in extremities’. We term such occurrences implicit entity mentions. While implicit entity mentions are common in many domains, resolving them is particularly valu- able in the clinical domain. Clinical documents are rich in information content that plays a cen- tral role in understanding patients’ health status and improving the quality of the delivered services. It is a common practice to employ computer assisted coding (CAC) solutions to assist expert “coders” in determining the unique identifier (e.g., ICD9 or ICD10) for each medical condition or combina- tion of conditions. These identifiers are impor- tant to unambiguously represent the medical con- ditions, to prepare the post-discharge plan, and to perform secondary data analysis tasks. A human coder reading the sentence ‘Patient has shortness of breath with reaccumulation of fluid in extremi- ties’ would generate the corresponding codes for en- tities ‘shortness of breath’ and ‘edema’. However, the solutions developed to perform entity recogni- tion in clinical documents (Aronson, 2006) (Fried- man et al., 1994) (Savova et al., …

Content truncated to 500 words