Detecting Hypoglycemia Incidents Reported in Patients' Secure Messages: Using Cost-Sensitive Learning and Oversampling to Reduce Data Imbalance.

BACKGROUND: Improper dosing of medications such as insulin can cause hypoglycemic episodes, which may lead to severe morbidity or even death. Although secure messaging was designed for exchanging nonurgent messages, patients sometimes report hypoglycemia events through secure messaging. Detecting these patient-reported adverse events may help alert clinical teams and enable early corrective actions to improve patient safety. OBJECTIVE: We aimed to develop a natural language processing system, called HypoDetect (Hypoglycemia Detector), to automatically identify hypoglycemia incidents reported in patients' secure messages. METHODS: An expert in public health annotated 3000 secure message threads between patients with diabetes and US Department of Veterans Affairs clinical teams as containing patient-reported hypoglycemia incidents or not. A physician independently annotated 100 threads randomly selected from this dataset to determine interannotator agreement. We used this dataset to develop and evaluate HypoDetect. HypoDetect incorporates 3 machine learning algorithms widely used for text classification: linear support vector machines, random forest, and logistic regression. We explored different learning features, including new knowledge-driven features. Because only 114 (3.80%) messages were annotated as positive, we investigated cost-sensitive learning and oversampling methods to mitigate the challenge of imbalanced data. RESULTS: The interannotator agreement was Cohen kappa=.976. Using cross-validation, logistic regression with cost-sensitive learning achieved the best performance (area under the receiver operating characteristic curve=0.954, sensitivity=0.693, specificity 0.974, F1 score=0.590). Cost-sensitive learning and the ensembled synthetic minority oversampling technique improved the sensitivity of the baseline systems substantially (by 0.123 to 0.728 absolute gains). Our results show that a variety of features contributed to the best performance of HypoDetect. CONCLUSIONS: Despite the challenge of data imbalance, HypoDetect achieved promising results for the task of detecting hypoglycemia incidents from secure messages. The system has a great potential to facilitate early detection and treatment of hypoglycemia.

A Natural Language Processing System That Links Medical Terms in Electronic Health Record Notes to Lay Definitions: System Development Using Physician Reviews.

BACKGROUND: Many health care systems now allow patients to access their electronic health record (EHR) notes online through patient portals. Medical jargon in EHR notes can confuse patients, which may interfere with potential benefits of patient access to EHR notes. OBJECTIVE: The aim of this study was to develop and evaluate the usability and content quality of NoteAid, a Web-based natural language processing system that links medical terms in EHR notes to lay definitions, that is, definitions easily understood by lay people. METHODS: NoteAid incorporates two core components: CoDeMed, a lexical resource of lay definitions for medical terms, and MedLink, a computational unit that links medical terms to lay definitions. We developed innovative computational methods, including an adapted distant supervision algorithm to prioritize medical terms important for EHR comprehension to facilitate the effort of building CoDeMed. Ten physician domain experts evaluated the user interface and content quality of NoteAid. The evaluation protocol included a cognitive walkthrough session and a postsession questionnaire. Physician feedback sessions were audio-recorded. We used standard content analysis methods to analyze qualitative data from these sessions. RESULTS: Physician feedback was mixed. Positive feedback on NoteAid included (1) Easy to use, (2) Good visual display, (3) Satisfactory system speed, and (4) Adequate lay definitions. Opportunities for improvement arising from evaluation sessions and feedback included (1) improving the display of definitions for partially matched terms, (2) including more medical terms in CoDeMed, (3) improving the handling of terms whose definitions vary depending on different contexts, and (4) standardizing the scope of definitions for medicines. On the basis of these results, we have improved NoteAid's user interface and a number of definitions, and added 4502 more definitions in CoDeMed. CONCLUSIONS: Physician evaluation yielded useful feedback for content validation and refinement of this innovative tool that has the potential to improve patient EHR comprehension and experience using patient portals. Future ongoing work will develop algorithms to handle ambiguous medical terms and test and evaluate NoteAid with patients.

Automated identification of eviction status from electronic health record notes.

OBJECTIVE: Evictions are important social and behavioral determinants of health. Evictions are associated with a cascade of negative events that can lead to unemployment, housing insecurity/homelessness, long-term poverty, and mental health problems. In this study, we developed a natural language processing system to automatically detect eviction status from electronic health record (EHR) notes. MATERIALS AND METHODS: We first defined eviction status (eviction presence and eviction period) and then annotated eviction status in 5000 EHR notes from the Veterans Health Administration (VHA). We developed a novel model, KIRESH, that has shown to substantially outperform other state-of-the-art models such as fine-tuning pretrained language models like BioBERT and Bio_ClinicalBERT. Moreover, we designed a novel prompt to further improve the model performance by using the intrinsic connection between the 2 subtasks of eviction presence and period prediction. Finally, we used the Temperature Scaling-based Calibration on our KIRESH-Prompt method to avoid overconfidence issues arising from the imbalance dataset. RESULTS: KIRESH-Prompt substantially outperformed strong baseline models including fine-tuning the Bio_ClinicalBERT model to achieve 0.74672 MCC, 0.71153 Macro-F1, and 0.83396 Micro-F1 in predicting eviction period and 0.66827 MCC, 0.62734 Macro-F1, and 0.7863 Micro-F1 in predicting eviction presence. We also conducted additional experiments on a benchmark social determinants of health (SBDH) dataset to demonstrate the generalizability of our methods. CONCLUSION AND FUTURE WORK: KIRESH-Prompt has substantially improved eviction status classification. We plan to deploy KIRESH-Prompt to the VHA EHRs as an eviction surveillance system to help address the US Veterans' housing insecurity.

An Investigation of the Representation of Social Determinants of Health in the UMLS.

Social Determinants of Health (SDOH) are the conditions in which people are born, live, work, and age1. Unified Medical Language System (UMLS) incorporates SDOH concepts2; but few have evaluated its coverage and quality. With 15,649 expert-annotated SDOH mentions from 3176 randomly selected electronic health record (EHR) notes, we found that 100% SDOH mentions can be mapped to at least one UMLS concept, indicating a good coverage of SDOH. However, we discovered a few challenges for the UMLS's representation of SDOH. Next, we developed a multi-step framework to identify SDOH concepts from UMLS, and a clinical BERT-based classification algorithm to assign each identified SDOH concept to one of the six general categories. Our multi-step framework extracted a total of 198, 677 SDOH concepts from the UMLS and the SDOH category classification system attained an accuracy of 91%. We also built EASE: an open-source tool to Extract SDOH from EHRs.

Learning to detect and understand drug discontinuation events from clinical narratives.

OBJECTIVE: Identifying drug discontinuation (DDC) events and understanding their reasons are important for medication management and drug safety surveillance. Structured data resources are often incomplete and lack reason information. In this article, we assessed the ability of natural language processing (NLP) systems to unlock DDC information from clinical narratives automatically. MATERIALS AND METHODS: We collected 1867 de-identified providers' notes from the University of Massachusetts Medical School hospital electronic health record system. Then 2 human experts chart reviewed those clinical notes to annotate DDC events and their reasons. Using the annotated data, we developed and evaluated NLP systems to automatically identify drug discontinuations and reasons at the sentence level using a novel semantic enrichment-based vector representation (SEVR) method for enhanced feature representation. RESULTS: Our SEVR-based NLP system achieved the best performance of 0.785 (AUC-ROC) for detecting discontinuation events and 0.745 (AUC-ROC) for identifying reasons when testing this highly imbalanced data, outperforming 2 state-of-the-art non-SEVR-based models. Compared with a rule-based baseline system for discontinuation detection, our system improved the sensitivity significantly (57.75% vs 18.31%, absolute value) while retaining a high specificity of 99.25%, leading to a significant improvement in AUC-ROC by 32.83% (absolute value). CONCLUSION: Experiments have shown that a high-performance NLP system can be developed to automatically identify DDCs and their reasons from providers' notes. The SEVR model effectively improved the system performance showing better generalization and robustness on unseen test data. Our work is an important step toward identifying reasons for drug discontinuation that will inform drug safety surveillance and pharmacovigilance.