JMIR Publications

ABSTRACT

Background:

Hypoglycemic events are common and potentially dangerous conditions among patients being treated for diabetes. Automatic detection of such events could improve patient care and is valuable in population studies. Electronic health records (EHRs) are valuable resources for the detection of such events.

Objective:

In this study, we aim to develop a deep-learning-based natural language processing system to automatically detect hypoglycemic events (HYPE) from EHR notes.

Methods:

Domain experts chart-reviewed 500 EHR notes of diabetes patients to annotate whether each sentence contained a hypoglycemic event or not. We used this annotated corpus to train and evaluate HYPE, the high-performance natural language processing system for hypoglycemia detection. We built and evaluated both a classical machine learning model (support vector machines) and state-of-the-art neural network models.

Results:

We found that neural network models outperformed the support vector machines (SVMs) model. The convolutional neural network model yielded the highest performance (precision = 0.96 ± 0.03, recall = 0.86 ± 0.03, F1 = 0.91 ± 0.03) in a 10-fold cross-validation setting.

Conclusions:

Despite the challenges posed by small and highly imbalanced data, our convolutional neural network–based HYPE system still achieved a high performance for hypoglycemia detection. HYPE could be used for EHR-based hypoglycemia surveillance and population studies in diabetes patients.