Classifying the Indication for Colonoscopy Procedures: A Comparison of NLP Approaches in a Diverse National Healthcare System.

In order to measure the level of utilization of colonoscopy procedures, identifying the primary indication for the procedure is required. Colonoscopies may be utilized not only for screening, but also for diagnostic or therapeutic purposes. To determine whether a colonoscopy was performed for screening, we created a natural language processing system to identify colonoscopy reports in the electronic medical record system and extract indications for the procedure. A rule-based model and three machine-learning models were created using 2,000 manually annotated clinical notes of patients cared for in the Department of Veterans Affairs. Performance of the models was measured and compared. Analysis of the models on a test set of 1,000 documents indicates that the rule-based system performance stays fairly constant as evaluated on training and testing sets. However, the machine learning model without feature selection showed significant decrease in performance. Therefore, rule-based classification system appears to be more robust than a machine-learning system in cases when no feature selection is performed.

Recognizing Questions and Answers in EMR Templates Using Natural Language Processing.

Templated boilerplate structures pose challenges to natural language processing (NLP) tools used for information extraction (IE). Routine error analyses while performing an IE task using Veterans Affairs (VA) medical records identified templates as an important cause of false positives. The baseline NLP pipeline (V3NLP) was adapted to recognize negation, questions and answers (QA) in various template types by adding a negation and slot:value identification annotator. The system was trained using a corpus of 975 documents developed as a reference standard for extracting psychosocial concepts. Iterative processing using the baseline tool and baseline+negation+QA revealed loss of numbers of concepts with a modest increase in true positives in several concept categories. Similar improvement was noted when the adapted V3NLP was used to process a random sample of 318,000 notes. We demonstrate the feasibility of adapting an NLP pipeline to recognize templates.

Automated identification of patients with a diagnosis of binge eating disorder from narrative electronic health records.

Binge eating disorder (BED) does not have an International Classification of Diseases, 9th or 10th edition code, but is included under 'eating disorder not otherwise specified' (EDNOS). This historical cohort study identified patients with clinician-diagnosed BED from electronic health records (EHR) in the Department of Veterans Affairs between 2000 and 2011 using natural language processing (NLP) and compared their characteristics to patients identified by EDNOS diagnosis codes. NLP identified 1487 BED patients with classification accuracy of 91.8% and sensitivity of 96.2% compared to human review. After applying study inclusion criteria, 525 patients had NLP-identified BED only, 1354 had EDNOS only, and 68 had both BED and EDNOS. Patient characteristics were similar between the groups. This is the first study to use NLP as a method to identify BED patients from EHR data and will allow further epidemiological study of patients with BED in systems with adequate clinical notes.

"Sitting on pins and needles": characterization of symptom descriptions in clinical notes".

Patients report their symptoms and subjective experiences in their own words. These expressions may be clinically meaningful yet are difficult to capture using automated methods. We annotated subjective symptom expressions in 750 clinical notes from the Veterans Affairs EHR. Within each document, subjective symptom expressions were compared to mentions of symptoms in clinical terms and to the assigned ICD-9-CM codes for the encounter. A total of 543 subjective symptom expressions were identified, of which 66.5% were categorized as mental/behavioral experiences and 33.5% somatic experiences. Only two subjective expressions were coded using ICD-9-CM. Subjective expressions were restated in semantically related clinical terms in 246 (45.3%) instances. Nearly one third (31%) of subjective expressions were not coded or restated in standard terminology. The results highlight the diversity of symptom descriptions and the opportunities to further develop natural language processing to extract symptom expressions that are unobtainable by other automated methods.

What a catch! traits that define good annotators.

Human annotation and chart review is an important process in biomedical informatics research, but which humans are best suited for the job? Domain expertise, such as medical or linguistic knowledge is desirable, but other factors may be equally important. The University of Utah has a group of 20+ reviewers with backgrounds in medicine and linguistics, and 10 key traits have surfaced in those best able to annotate quickly and with high quality. To identify reviewers with these key traits, we created a hiring process that includes interviewing candidates, testing their medical and linguistic knowledge, and having them complete an annotation exercise on realistic medical text. Each step is designed to assess the key traits and allow the investigator to choose the skill set required for each project.

Evaluating the state of the art in coreference resolution for electronic medical records.

BACKGROUND: The fifth i2b2/VA Workshop on Natural Language Processing Challenges for Clinical Records conducted a systematic review on resolution of noun phrase coreference in medical records. Informatics for Integrating Biology and the Bedside (i2b2) and the Veterans Affair (VA) Consortium for Healthcare Informatics Research (CHIR) partnered to organize the coreference challenge. They provided the research community with two corpora of medical records for the development and evaluation of the coreference resolution systems. These corpora contained various record types (ie, discharge summaries, pathology reports) from multiple institutions. METHODS: The coreference challenge provided the community with two annotated ground truth corpora and evaluated systems on coreference resolution in two ways: first, it evaluated systems for their ability to identify mentions of concepts and to link together those mentions. Second, it evaluated the ability of the systems to link together ground truth mentions that refer to the same entity. Twenty teams representing 29 organizations and nine countries participated in the coreference challenge. RESULTS: The teams' system submissions showed that machine-learning and rule-based approaches worked best when augmented with external knowledge sources and coreference clues extracted from document structure. The systems performed better in coreference resolution when provided with ground truth mentions. Overall, the systems struggled in solving coreference resolution for cases that required domain knowledge.