A System for Predicting ICD-10-PCS Codes from Electronic Health Records

Abstract

Medical coding is a process of classifying health records according to standard code sets representing procedures and diagnoses. It is an integral part of health care in the U.S., and the high costs it incurs have prompted adoption of natural language processing techniques for automatic generation of these codes from the clinical narrative contained in electronic health records. The need for effective auto-coding methods becomes even greater with the impending adoption of ICD-10, a code inventory of greater complexity than the currently used code sets. This paper presents a system that predicts ICD-10 procedure codes from the clinical narrative using several levels of abstraction. First, partial hierarchical classification is used to identify potentially relevant concepts and codes. Then, for each of these concepts we estimate the confidence that it appears in a procedure code for that document. Finally, confidence values for the candidate codes are estimated using features derived from concept confidence scores. The concept models can be trained on data with ICD-9 codes to supplement sparse ICD-10 training resources. Evaluation on held-out data shows promising results.

ICD-10 Coding of Spanish Electronic Discharge Summaries: An Extreme Classification Problem

“…As an alternative to biomedical dictionaries, other authors choose to reduce bias and extend collections by transforming other data sets. Subotin et al use the General Equivalence Mappings (GEMS 2 ) between ICD-10 and ICD-9 to supplement the small size of the training corpus through reports annotated with ICD-9 [8]. In turn, Almagro et al explore the application of Machine Translation techniques to expand the data set with foreign resources [9].…”

Towards automated clinical coding

“…Several studies adopt model architectures which reflect this structure. One approach trains a binary SVM for each node in an ontology, with each classifier learning only from training examples classed as positive by its parent classifier [34,35,36,37,38]. A framework has been described for feedforward neural network training which is regularised so as to incorporate tree-based priors derived from disease ontologies [39].…”

Aspiring to Unintended Consequences of Natural Language Processing: A Review of Recent Developments in Clinical and Consumer-Generated Text Processing

“…Their feature set is a simple bag-of-words from clinical notes, but the classification itself leverages the hierarchical nature of the ICD-9 tree. With the adoption of ICD-10 coding, Subotin and Davis proposed a diagnosis code assignment method, which also considers a bag-of-words approach, but combines a series of assignments based in part on the structure of the ICD-10 classification [58]. Their experiments on a corpus of 28,000 patient records show promising results for this new and complex terminology.…”