OmniTFT: Omni Target Forecasting for Vital Signs and Laboratory Result Trajectories in Multi Center ICU Data

2025-11-23

AI Summary

ICU multivariate time-series forecasting faces challenges including high noise and rapid fluctuations in vital signs, as well as substantial missingness, measurement delays, and device-specific biases in laboratory data. To address these issues, we propose a joint prediction framework tailored for clinical heterogeneous data: (1) sliding-window balanced sampling mitigates temporal imbalance; (2) a frequency-aware embedding compression module unifies modeling of high-frequency vital signs and sparse lab results; and (3) a hierarchical variable selection and effect-aligned attention mechanism enables physiology-informed feature fusion and calibration within the Temporal Fusion Transformer (TFT) architecture. Evaluated on three multicenter datasets—MIMIC-III, MIMIC-IV, and eICU—the model achieves significant improvements in prediction accuracy (average MAE reduced by 12.7%), exhibits attention patterns consistent with established pathophysiological mechanisms, and demonstrates strong cross-institutional generalizability and clinical interpretability.

Technology Category

• Machine Learning: Time-Series/Data Streams
• Computer Vision: Multi-modal Vision
• Intelligent Robots: Multimodal Perception & Sensor Fusion

Application Category

• Graph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Accurate multivariate time-series prediction of vital signs and laboratory results is crucial for early intervention and precision medicine in intensive care units (ICUs). However, vital signs are often noisy and exhibit rapid fluctuations, while laboratory tests suffer from missing values, measurement lags, and device-specific bias, making integrative forecasting highly challenging. To address these issues, we propose OmniTFT, a deep learning framework that jointly learns and forecasts high-frequency vital signs and sparsely sampled laboratory results based on the Temporal Fusion Transformer (TFT). Specifically, OmniTFT implements four novel strategies to enhance performance: sliding window equalized sampling to balance physiological states, frequency-aware embedding shrinkage to stabilize rare-class representations, hierarchical variable selection to guide model attention toward informative feature clusters, and influence-aligned attention calibration to enhance robustness during abrupt physiological changes. By reducing the reliance on target-specific architectures and extensive feature engineering, OmniTFT enables unified modeling of multiple heterogeneous clinical targets while preserving cross-institutional generalizability. Across forecasting tasks, OmniTFT achieves substantial performance improvement for both vital signs and laboratory results on the MIMIC-III, MIMIC-IV, and eICU datasets. Its attention patterns are interpretable and consistent with known pathophysiology, underscoring its potential utility for quantitative decision support in clinical care.

Problem

• Forecasting noisy vital signs and sparse lab results in ICU data
• Addressing missing values and device bias in clinical time-series
• Developing unified model for heterogeneous targets across multiple institutions

Innovation

• Sliding window equalized sampling balances physiological states
• Frequency-aware embedding shrinkage stabilizes rare-class representations
• Hierarchical variable selection guides attention to informative features