Time-to-Event Transformer to Capture Timing Attention of Events in EHR Time Series

2026-02-11

AI Summary

Current AI models struggle to effectively capture the temporal ordering of events and individualized timing patterns in electronic health records (EHRs), limiting causal inference and prediction in precision medicine. To address this, this work proposes LITT, a novel Timing-Transformer architecture that, for the first time, treats event timing as a computable dimension. LITT introduces a virtual “relative time axis” to temporally align event sequences and incorporates a timing-centric attention mechanism. By integrating relative time encoding, survival analysis, and temporal alignment, the method accurately predicts the onset time of cardiotoxicity-related cardiac events in real-world EHR data from 3,276 breast cancer patients and significantly outperforms existing benchmarks and state-of-the-art survival analysis approaches across multiple public datasets.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Reasoning under Uncertainty: Causality
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data

Application Category

• Search and Retrieval-Augmented AI: Large language models for search
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

Abstract

Automatically discovering personalized sequential events from large-scale time-series data is crucial for enabling precision medicine in clinical research, yet it remains a formidable challenge even for contemporary AI models. For example, while transformers capture rich associations, they are mostly agnostic to event timing and ordering, thereby bypassing potential causal reasoning. Intuitively, we need a method capable of evaluating the"degree of alignment"among patient-specific trajectories and identifying their shared patterns, i.e., the significant events in a consistent sequence. This necessitates treating timing as a true \emph{computable} dimension, allowing models to assign ``relative timestamps''to candidate events beyond their observed physical times. In this work, we introduce LITT, a novel Timing-Transformer architecture that enables temporary alignment of sequential events on a virtual ``relative timeline'', thereby enabling \emph{event-timing-focused attention} and personalized interpretations of clinical trajectories. Its interpretability and effectiveness are validated on real-world longitudinal EHR data from 3,276 breast cancer patients to predict the onset timing of cardiotoxicity-induced heart disease. Furthermore, LITT outperforms both the benchmark and state-of-the-art survival analysis methods on public datasets, positioning it as a significant step forward for precision medicine in clinical AI.

Problem

• Time-to-Event
• Event Timing
• Electronic Health Records
• Temporal Alignment
• Precision Medicine

Innovation

• Time-to-Event Transformer
• relative timeline
• timing-aware attention
• personalized clinical trajectories
• survival analysis