EHR phenotyping via jointly embedding medical concepts and words into a unified vector space

Abstract

BackgroundThere has been an increasing interest in learning low-dimensional vector representations of medical concepts from Electronic Health Records (EHRs). Vector representations of medical concepts facilitate exploratory analysis and predictive modeling of EHR data to gain insights about the patterns of care and health outcomes. EHRs contain structured data such as diagnostic codes and laboratory tests, as well as unstructured free text data in form of clinical notes, which provide more detail about condition and treatment of patients.MethodsIn this work, we propose a method that jointly learns vector representations of medical concepts and words. This is achieved by a novel learning scheme based on the word2vec model. Our model learns those relationships by integrating clinical notes and sets of accompanying medical codes and by defining joint contexts for each observed word and medical code.ResultsIn our experiments, we learned joint representations using MIMIC-III data. Using the learned representations of words and medical codes, we evaluated phenotypes for 6 diseases discovered by our and baseline method. The experimental results show that for each of the 6 diseases our method finds highly relevant words. We also show that our representations can be very useful when predicting the reason for the next visit.ConclusionsThe jointly learned representations of medical concepts and words capture not only similarity between codes or words themselves, but also similarity between codes and words. They can be used to extract phenotypes of different diseases. The representations learned by the joint model are also useful for construction of patient features.

Deep Learning for Electronic Health Records Analytics

“…Use Case 2 (Embedding Medical Concepts and Words Into a Unified Vector Space [62]): Most of the studies who tried to leverage EHR data for patient's phenotyping used the embedding of medical codes like the ICD9 and fed the resulting vectors to a neural network to establish diseases phenotypes or to predict a clinical adverse event [63]- [66]. Other approaches have tried to embed the extracted medical codes and accompanying words separately.…”

A Systematic Review of Natural Language Processing for Knowledge Management in Healthcare

“…Knowledge management process of externalization is also supported by the NLP through converting the tacit knowledge stored in medical database systems such as the unified medical language systems and converting it to explicit information [4], [13]. The NLP supports knowledge management in information capture and extraction through the NLP systems such as MedLEE proposed by [15], the general architecture for text engineering (GATE) by [16], and [27], the Word2Vec by [23], the conditional random field by [14], the Unstructured Information Management Architecture (UIMMA) [26].…”

Lexicon Development for COVID-19-related Concepts Using Open-source Word Embedding Sources: An Intrinsic and Extrinsic Evaluation

“…Word embeddings have advanced the state of the art for many intrinsic natural language processing subtasks (ie, reading comprehension [ 22 ], natural language inference [ 23 ], text summarization [ 24 ], vocabulary development [ 8 ], and document classification [ 25 ]). An extrinsic or summative evaluation of clinical word embeddings can involve evaluating the performance of machine learning models by using word embeddings to complete a biomedical research task or clinical operation such as patient phenotyping [ 26 , 27 ], patient fall prediction [ 25 ], and patient hospital readmission prediction [ 28 ].…”