Key result
EHR-derived Word2vec embeddings improve 30-day HF readmission prediction over baseline one-hot encoding.
Why the study?
Heart failure is associated with high rates of unplanned readmissions, and existing models to predict these readmissions have shown suboptimal performance.
Do EHR feature embedding approaches (word2vec on medical codes/CUIs and BioClinicalBERT on concept descriptions) improve the prediction of 30-day hospital readmission compared to one-hot encoding in patients with heart failure?
Observational (n=21,031)
No
Do EHR feature embedding approaches (word2vec on medical codes/CUIs and BioClinicalBERT on concept descriptions) improve the prediction of 30-day hospital readmission compared to one-hot encoding in patients with heart failure?
Word2vec embeddings trained directly on structured EHR medical codes and CUIs outperform both one-hot encoding and pre-trained BioClinicalBERT text embeddings for predicting 30-day hospital readmission in heart failure patients (AUROC 0.65 vs 0.54 and 0.59).
May refine HF readmission risk models; hypothesis-generating and requires prospective validation before clinical use.
Background: Heart failure (HF) is a public health concern with a wider impact on quality of life and cost of care. One of the major challenges in HF is the higher rate of unplanned readmissions and suboptimal performance of models to predict the readmissions. Hence, in this study, we implemented embeddings-based approaches to generate features for improving model performance. Objective: The objective of this study was to evaluate and compare the effectiveness of different feature embedding approaches for improving the prediction of unplanned readmissions in patients with heart failure. Methods: We compared three embedding approaches including word2vec on terminology codes and concept unique identifier (CUIs) and BERT on descriptive text of concept with baseline (one hot-encoding). We compared area under the receiver operating characteristic (AUROC) and F1-scores for the logistic regression, eXtream gradient-boosting (XGBoost) and artificial neural network (ANN) models using these embedding approaches. The model was tested on the heart failure cohort (N=21,031) identified using least restrictive phenotyping methods from MIMIC-IV dataset. Results: We found that the embedding approaches significantly improved the performance of the prediction models. The XGBoost performed better for all approaches. The word2vec embeddings (0.65) trained on the dataset outperformed embeddings from pre-trained BERT model (0.59) using descriptive text. Conclusions: Embedding methods, particularly word2vec trained on electronic health record data, can better discriminate HF readmission cases compared to both one-hot encoding and pre-trained BERT embeddings on concept descriptions making it a viable approach of automation feature selection. The observed AUROC improvement (0.65 vs 0.54) may support more effective risk stratification and targeted clinical interventions.
No takes yet. Share an insight, caveat, or question.
Shakya et al. (2025) conducted an observational in Heart Failure (n=21,031). Word2vec embeddings on CUIs with XGBoost vs. One-hot encoding (baseline) was evaluated on AUROC for 30-day readmission. Word2vec embeddings trained on electronic health record data improved the prediction of 30-day heart failure readmissions, achieving an AUROC of 0.65 compared to 0.54 for baseline one-hot encoding.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: