Dynamic Survival Transformers for Causal Inference with Electronic Health Records
This addresses the need for more accurate causal inference in medical research using EHRs, though it appears incremental as it builds on existing transformer methods for survival analysis.
The paper tackled the problem of causal survival analysis in medicine by introducing the Dynamic Survival Transformer (DynST), which uses time-varying EHR data to estimate treatment effects on survival outcomes, showing it outperforms alternative models in predictive and causal estimation on a semi-synthetic MIMIC-III dataset.
In medicine, researchers often seek to infer the effects of a given treatment on patients' outcomes. However, the standard methods for causal survival analysis make simplistic assumptions about the data-generating process and cannot capture complex interactions among patient covariates. We introduce the Dynamic Survival Transformer (DynST), a deep survival model that trains on electronic health records (EHRs). Unlike previous transformers used in survival analysis, DynST can make use of time-varying information to predict evolving survival probabilities. We derive a semi-synthetic EHR dataset from MIMIC-III to show that DynST can accurately estimate the causal effect of a treatment intervention on restricted mean survival time (RMST). We demonstrate that DynST achieves better predictive and causal estimation than two alternative models.