SigBERT: Combining Narrative Medical Reports and Rough Path Signature Theory for Survival Risk Estimation in Oncology

TL;DR

SigBERT introduces a novel framework combining narrative medical reports and rough path signature theory to improve survival risk estimation in oncology, effectively capturing complex temporal dynamics from textual data.

Contribution

The paper presents SigBERT, a new method that leverages signature extraction from rough path theory to process sequential medical reports for survival analysis, outperforming existing approaches.

Findings

Achieved a C-index of 0.75 on real-world oncology data.

Effectively captures temporal dynamics from narrative reports.

Enhances survival risk estimation accuracy.

Abstract

Electronic medical reports (EHR) contain a vast amount of information that can be leveraged for machine learning applications in healthcare. However, existing survival analysis methods often struggle to effectively handle the complexity of textual data, particularly in its sequential form. Here, we propose SigBERT, an innovative temporal survival analysis framework designed to efficiently process a large number of clinical reports per patient. SigBERT processes timestamped medical reports by extracting and averaging word embeddings into sentence embeddings. To capture temporal dynamics from the time series of sentence embedding coordinates, we apply signature extraction from rough path theory to derive geometric features for each patient, which significantly enhance survival model performance by capturing complex temporal dynamics. These features are then integrated into a LASSO-penalized Cox model to estimate patient-specific risk scores. The model was trained and evaluated on a real-world oncology dataset from the L\'eon B\'erard Center corpus, with a C-index score of 0.75 (sd 0.014) on the independent test cohort. SigBERT integrates sequential medical data to enhance risk estimation, advancing narrative-based survival analysis.