An attention based deep learning model of clinical events in the intensive care unit.
Deepak A KajiJohn R ZechJun S KimSamuel K ChoNeha S DangayachAnthony B CostaEric K OermannPublished in: PloS one (2019)
This study trained long short-term memory (LSTM) recurrent neural networks (RNNs) incorporating an attention mechanism to predict daily sepsis, myocardial infarction (MI), and vancomycin antibiotic administration over two week patient ICU courses in the MIMIC-III dataset. These models achieved next-day predictive AUC of 0.876 for sepsis, 0.823 for MI, and 0.833 for vancomycin administration. Attention maps built from these models highlighted those times when input variables most influenced predictions and could provide a degree of interpretability to clinicians. These models appeared to attend to variables that were proxies for clinician decision-making, demonstrating a challenge of using flexible deep learning approaches trained with EHR data to build clinical decision support. While continued development and refinement is needed, we believe that such models could one day prove useful in reducing information overload for ICU physicians by providing needed clinical decision support for a variety of clinically important tasks.
Keyphrases
- clinical decision support
- electronic health record
- working memory
- deep learning
- intensive care unit
- neural network
- decision making
- acute kidney injury
- primary care
- methicillin resistant staphylococcus aureus
- heart failure
- artificial intelligence
- healthcare
- resistance training
- machine learning
- mechanical ventilation
- clinical trial
- physical activity
- convolutional neural network
- palliative care
- left ventricular
- big data
- high intensity
- study protocol
- extracorporeal membrane oxygenation
- health information
- placebo controlled