Smoking is an important variable for clinical research, but there are few studies regarding automatic obtainment of smoking classification from unstructured bilingual electronic health records (EHR). We aim to develop an algorithm to classify smoking status based on unstructured EHRs using natural language processing (NLP). With acronym replacement and Python package Soynlp, we normalize 4711 bilingual clinical notes. Each EHR notes was classified into 4 categories: current smokers, past smokers, never smokers, and unknown. Subsequently, SPPMI (Shifted Positive Point Mutual Information) is used to vectorize words in the notes. By calculating cosine similarity between these word vectors, keywords denoting the same smoking status are identified. Compared to other keyword extraction methods (word co-occurrence-, PMI-, and NPMI-based methods), our proposed approach improves keyword extraction precision by as much as 20.0%. These extracted keywords are used in classifying 4 smoking statuses from our bilingual EHRs. Given an identical SVM classifier, the F1 score is improved by as much as 1.8% compared to those of the unigram and bigram Bag of Words. Our study shows the potential of SPPMI in classifying smoking status from bilingual, unstructured EHRs. Our current findings show how smoking information can be easily acquired for clinical practice and research.

As a key modifiable risk factor, alcohol consumption is clinically crucial information that allows medical professionals to further understand their patients’ medical conditions and suggest appropriate lifestyle modifying interventions. However, identifying alcohol-related information from unstructured free-text clinical notes is often challenging. Not only are the formats of the notes inconsistent, but they also include a massive amount of non-alcohol-related information. Furthermore, for medical institutions outside of English-speaking countries, these clinical notes contain both a mixture of English and local languages, inducing additional difficulty in the extraction. Thanks to the increasing availability of electronic medical record (EMR), several previous works explored the idea of using natural language processing (NLP) to train machine learning models that automatically identify alcohol-related information from unstructured clinical notes. However, all these previous works are limited to English clinical notes, thereby able to leverage various large-scale external ontologies during the text preprocessing. Furthermore, they rely on simple NLP techniques such as the bag-of-words models that suffer from high dimensionality and out-of-vocabulary issues. Addressing these issues, we adopt fine-tuning multilingual transformers. By leveraging their linguistically rich contextual information learned during their pre-training, we are able to extract alcohol-related information from unstructured clinical notes without preprocessing the clinical notes on any external ontologies. Furthermore, our work is the first to explore the use of transformers in bilingual clinical notes to extract alcohol-related information. Even with minimal text preprocessing, we achieve extraction accuracy of 84.70% in terms of macro F-1 score.