JMIR medical informatics · 2026
Predicting Laboratory Test Ordering in Emergency Departments Using Integrated Structured and Unstructured Electronic Health Records: Machine Learning Study.
СтатьяПоддержка исследования: вне правительства СШАНазвание и аннотация: Перевод в очереди
Обработка статьи · этапы, ошибки и управление
Загрузка состояния…
Ошибки и ограничения
История · было → стало
Состояние обновляется каждые 5 секунд, пока блок открыт. Процент показывается только там, где известен объём работы. Пауза останавливает продолжение этапа; текущий запрос к источнику может завершиться.
Zhang Xingyu, Ling Haipeng, Zhang Xin, Zhang Anao
Аннотация
BACKGROUND: Laboratory testing is a cornerstone of diagnostic decision-making in emergency departments (EDs), yet its overuse contributes substantially to unnecessary health care costs and inefficiencies. Predictive approaches that leverage electronic health record data may help optimize and guide more appropriate test use.
OBJECTIVE: This study aims to develop and evaluate machine learning models that predict laboratory test use during ED visits by integrating structured clinical data and unstructured text from electronic health records.
METHODS: We analyzed 13,115 adult ED visits from the 2021 National Hospital Ambulatory Medical Care Survey-Emergency Department dataset. Structured predictors included demographics, vital signs, insurance status, and medical history, while unstructured data from chief complaints and injury descriptions were encoded using Bidirectional Encoder Representations from Transformers-based embeddings. Four model configurations were developed: structured-only, unstructured-only, combined structured and unstructured data, and an ensemble (mean probability) approach. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC).
RESULTS: The combined model achieved the highest predictive performance (AUC=0.83), outperforming both the structured-only model (AUC=0.78) and the unstructured-only model (AUC=0.74). The ensemble model also performed well but did not exceed the combined model. Key predictors of laboratory testing included older age, ambulance arrival, abnormal vital signs, and chronic comorbidities, whereas injury-related visits were associated with a lower likelihood of testing.
CONCLUSIONS: Integrating structured and unstructured electronic health record data improves the prediction of laboratory test use in ED settings. These findings support the development of data-driven clinical decision support tools aimed at enhancing diagnostic efficiency and reducing unnecessary testing.
OBJECTIVE: This study aims to develop and evaluate machine learning models that predict laboratory test use during ED visits by integrating structured clinical data and unstructured text from electronic health records.
METHODS: We analyzed 13,115 adult ED visits from the 2021 National Hospital Ambulatory Medical Care Survey-Emergency Department dataset. Structured predictors included demographics, vital signs, insurance status, and medical history, while unstructured data from chief complaints and injury descriptions were encoded using Bidirectional Encoder Representations from Transformers-based embeddings. Four model configurations were developed: structured-only, unstructured-only, combined structured and unstructured data, and an ensemble (mean probability) approach. Model performance was evaluated using the area under the receiver operating characteristic curve (AUC).
RESULTS: The combined model achieved the highest predictive performance (AUC=0.83), outperforming both the structured-only model (AUC=0.78) and the unstructured-only model (AUC=0.74). The ensemble model also performed well but did not exceed the combined model. Key predictors of laboratory testing included older age, ambulance arrival, abnormal vital signs, and chronic comorbidities, whereas injury-related visits were associated with a lower likelihood of testing.
CONCLUSIONS: Integrating structured and unstructured electronic health record data improves the prediction of laboratory test use in ED settings. These findings support the development of data-driven clinical decision support tools aimed at enhancing diagnostic efficiency and reducing unnecessary testing.