Several natural language processing (NLP) pipelines have been reported in recent years that utilize combinations of essential components including: tokenization, part-of-speech tagging, named entity recognition, negation and mapping to Unified Medical Language System (UMLS) ontologies. Although International Classification of Diseases (ICD) codes are commonly used for phenotyping patients based on EHR, coder errors, such as misattribution, unbundling, and upcoding, result in low sensitivity and specificity for retrieval of reliable clinical information. However, the presence or absence of AMS requires extraction of the information from the providers’ clinical text notes. Most of the other patient characteristics needed for PESI can be extracted from coded EHR data, e.g. Based on these reports, and for the purpose of this experiment, we defined AMS as the presence of any of the following symptoms: disorientation, confusion, somnolence, lethargy, stupor, syncope or coma. According to the PESI guideline, the presence of AMS significantly increases the risk of post-pulmonary embolism mortality. The Pulmonary Embolism Severity Index (PESI) is a risk stratification guideline that helps clinicians assess patients with pulmonary embolism and determine the necessary practice guidelines for treatment and follow-up care. Pulmonary embolism should be considered during the evaluation of patients with syncope. Our objective is to automate the detection of altered mental status (AMS) in ED provider notes for the ultimate use in clinical decision support. This study was motivated by the need for the assessment of mental status during the evaluation and risk stratification of patients with pulmonary embolism in the emergency department (ED).
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |