Explainable Prediction of Medical Codes from Clinical Text

Proceedings of NAACL-HLT 2018, pages 1101–1111 New Orleans, Louisiana, June 1 - 6, 2018. c⃝2018 Association for Computational Linguistics Explainable Prediction of Medical Codes from Clinical Text James Mullenbach, Sarah Wiegreffe, Jon Duke, Jimeng Sun, Jacob Eisenstein Georgia Institute of Technology {jmullenbach3, swiegreffe6, jon.duke}@gatech.edu jsun@cc.gatech.edu, jacobe@gatech.edu Abstract Clinical notes are text documents that are cre- ated by clinicians for each patient encounter. They are typically accompanied by medical codes, which describe the diagnosis and treat- ment. Annotating these codes is labor inten- sive and error prone; furthermore, the connec- tion between the codes and the text is not anno- tated, obscuring the reasons and details behind specific diagnoses and treatments. We present an attentional convolutional network that pre- dicts medical codes from clinical text. Our method aggregates information across the doc- ument using a convolutional neural network, and uses an attention mechanism to select the most relevant segments for each of the thou- sands of possible codes. The method is ac- curate, achieving precision@8 of 0.71 and a Micro-F1 of 0.54, which are both better than the prior state of the art. Furthermore, through an interpretability evaluation by a physician, we show that the attention mechanism identi- fies meaningful explanations for each code as- signment. 1 Introduction Clinical notes are free text narratives generated by clinicians during patient encounters. They are typically accompanied by a set of metadata codes from the International Classification of Diseases (ICD), which present a standardized way of in- dicating diagnoses and procedures that were per- formed during the encounter. ICD codes have a variety of uses, ranging from billing to predic- tive modeling of patient state (Choi et al., 2016; Ranganath et al., 2015; Denny et al., 2010; Avati et al., 2017). Because manual coding is time- consuming and error-prone, automatic coding has been studied since at least the 1990s (de Lima et al., 1998). The task is difficult for two main reasons. First, the label space is very high-dimensional, with over 15,000 codes in the ICD-9 taxonomy, and over 140,000 codes combined in the newer ICD-10-CM and ICD-10-PCS taxonomies (World Health Organization, 2016). Second, clinical text includes irrelevant information, misspellings and non-standard abbreviations, and a large medical vocabulary. These features combine to make the prediction of ICD codes from clinical notes an es- pecially difficult task, for computers and human coders alike (Birman-Deych et al., 2005). In this application paper, we develop convolu- tional neural network (CNN)-based methods for automatic ICD code assignment based on text dis- charge summaries from intensive care unit (ICU) stays. To better adapt to the multi-label setting, we employ a per-label attention mechanism, which allows our model to learn distinct document rep- resentations for each label. We call our method Convolutional Attention for Multi-Label classifi- cation (CAML). Our model design is motivated by the conjecture that important information cor- related with a code’s presence may be contained in short snippets of text which could be anywhere in the document, and that these snippets likely differ for different labels. To …

Content truncated to 500 words