TAPER: Time-Aware Patient EHR Representation
This addresses the problem of improving predictive healthcare outcomes for patients by providing a more accurate representation of EHR data, though it is incremental as it builds on existing transformer and BERT methods.
The paper tackled the challenge of learning effective representations from irregular, multi-modal electronic health records by using transformer networks and BERT to embed data into a unified vector, achieving superior performance on mortality, readmission, and length-of-stay tasks using the MIMIC-III ICU dataset.
Effective representation learning of electronic health records is a challenging task and is becoming more important as the availability of such data is becoming pervasive. The data contained in these records are irregular and contain multiple modalities such as notes, and medical codes. They are preempted by medical conditions the patient may have, and are typically jotted down by medical staff. Accompanying codes are notes containing valuable information about patients beyond the structured information contained in electronic health records. We use transformer networks and the recently proposed BERT language model to embed these data streams into a unified vector representation. The presented approach effectively encodes a patient's visit data into a single distributed representation, which can be used for downstream tasks. Our model demonstrates superior performance and generalization on mortality, readmission and length of stay tasks using the publicly available MIMIC-III ICU dataset. Code avaialble at https://github.com/sajaddarabi/TAPER-EHR