Early Risk Prediction of Pediatric Cardiac Arrest from Electronic Health Records via Multimodal Fused Transformer

2025-02-11

AI Summary

Early warning of cardiac arrest (CA) in pediatric intensive care remains a critical clinical challenge. To address this, we propose PedCA-FT, a novel multimodal fusion Transformer model specifically designed for pediatric CA risk prediction. PedCA-FT introduces a dual-view architecture that separately encodes structured longitudinal electronic health record (EHR) time-series data and unstructured clinical narrative text, integrating them via cross-modal attention to capture dynamic interactions and temporal evolution among high-dimensional risk factors. The model further supports clinically interpretable risk attribution analysis. Evaluated on the real-world CHO-A-CICU pediatric cohort, PedCA-FT achieves statistically significant improvements over ten state-of-the-art AI baselines across five key metrics: AUC, sensitivity, specificity, F1-score, and calibration. Moreover, it identifies clinically meaningful risk factors, delivering a deployable, interpretable, and clinically actionable early-warning system for pediatric CA.

Technology Category

• Intelligent Robots: Multimodal Perception & Sensor Fusion
• Computer Vision: Multi-modal Vision
• Reasoning under Uncertainty: Causality

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Retrieval-Augmented Generation (RAG) and multi-modal RAG
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Early prediction of pediatric cardiac arrest (CA) is critical for timely intervention in high-risk intensive care settings. We introduce PedCA-FT, a novel transformer-based framework that fuses tabular view of EHR with the derived textual view of EHR to fully unleash the interactions of high-dimensional risk factors and their dynamics. By employing dedicated transformer modules for each modality view, PedCA-FT captures complex temporal and contextual patterns to produce robust CA risk estimates. Evaluated on a curated pediatric cohort from the CHOA-CICU database, our approach outperforms ten other artificial intelligence models across five key performance metrics and identifies clinically meaningful risk factors. These findings underscore the potential of multimodal fusion techniques to enhance early CA detection and improve patient care.

Problem

• Predict pediatric cardiac arrest early
• Fuse EHR data for risk analysis
• Enhance detection with multimodal transformers

Innovation

• Transformer-based multimodal fusion
• Captures temporal and contextual patterns
• Outperforms ten AI models