Deep Survival Analysis for Competing Risk Modeling with Functional Covariates and Missing Data Imputation

2025-09-29

AI Summary

This paper addresses the challenge of competing-risks survival analysis in intensive care, where functional covariates (e.g., time-varying physiological signals) and missing data co-occur. We propose FCRN, an end-to-end deep learning framework that jointly models functional covariates via basis-function-based micro-networks, imputes missing values through a gradient-driven collaborative imputation mechanism, and estimates event-specific discrete-time hazards—all within a unified architecture optimized end-to-end. Evaluated on real-world ICU datasets (MIMIC-IV, Cleveland Clinic) and multiple synthetic benchmarks, FCRN significantly outperforms random survival forests and conventional competing-risks models in predictive accuracy, demonstrating robustness under high missingness rates and multi-event settings. Its core contribution is the first integrated learning paradigm that simultaneously learns functional representations, performs missing-data imputation, and estimates competing risks—enabling synergistic optimization across all three tasks.

Technology Category

• Machine Learning: Probabilistic Circuits and Graphical Models
• Reasoning under Uncertainty: Relational Probabilistic Models
• Computer Vision: Learning & Optimization for CV

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Graph Algorithms and Modeling for the Web: Algorithms and analysis for incomplete, noisy, or partially observed Web-related graphs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

We introduce the Functional Competing Risk Net (FCRN), a unified deep-learning framework for discrete-time survival analysis under competing risks, which seamlessly integrates functional covariates and handles missing data within an end-to-end model. By combining a micro-network Basis Layer for functional data representation with a gradient-based imputation module, FCRN simultaneously learns to impute missing values and predict event-specific hazards. Evaluated on multiple simulated datasets and a real-world ICU case study using the MIMIC-IV and Cleveland Clinic datasets, FCRN demonstrates substantial improvements in prediction accuracy over random survival forests and traditional competing risks models. This approach advances prognostic modeling in critical care by more effectively capturing dynamic risk factors and static predictors while accommodating irregular and incomplete data.

Problem

• Modeling competing risks with functional covariates and missing data
• Simultaneously imputing missing values and predicting event-specific hazards
• Improving prediction accuracy over traditional survival analysis methods

Innovation

• Deep learning framework for competing risk survival analysis
• Integrates functional covariates and handles missing data
• Combines basis layer with gradient-based imputation module