ClinicalBERT: Modeling Clinical Notes and Predicting Hospital Readmission
About
Clinical notes contain information about patients that goes beyond structured data like lab values and medications. However, clinical notes have been underused relative to structured data, because notes are high-dimensional and sparse. This work develops and evaluates representations of clinical notes using bidirectional transformers (ClinicalBERT). ClinicalBERT uncovers high-quality relationships between medical concepts as judged by humans. ClinicalBert outperforms baselines on 30-day hospital readmission prediction using both discharge summaries and the first few days of notes in the intensive care unit. Code and model parameters are available.
Kexin Huang, Jaan Altosaar, Rajesh Ranganath• 2019
Related benchmarks
| Task | Dataset | Result | Rank | |
|---|---|---|---|---|
| Procedures Prediction | MIMIC III Admission Notes v1.4 (test) | Macro AUROC86.15 | 14 | |
| Diagnoses Prediction | MIMIC III Admission Notes v1.4 (test) | Macro AUROC0.8199 | 14 | |
| In-hospital mortality prediction | MIMIC III Admission Notes v1.4 (test) | Macro AUROC82.2 | 10 | |
| Length-of-Stay Prediction | MIMIC III Admission Notes v1.4 (test) | macro-averaged AUROC71.14 | 10 | |
| Radiology Report Summarization | Radiology Report Summarization dataset (test) | GFLOPS50 | 8 | |
| Stroke outcome prediction | Stroke Outcome Prediction Dataset | MAE1.24 | 5 |
Showing 6 of 6 rows