What question did this study set out to answer?

This research aims to develop a novel framework for predicting myocardial infarction using multiple data modalities.

March 6, 2026IEEE Journal of Biomedical and Health Informatics1 citations

Graph-Informed and FiLM-Enhanced Multimodal Fusion for Myocardial Infarction Prediction

XXXiantong Xiang LGLongxiao Gao YLYuansheng Liu

Key Result

GFM-MIP, integrating 12-lead ECG signals, images, and labs, outperformed state-of-the-art methods for myocardial infarction prediction across multiple datasets.

Key Points

This research aims to develop a novel framework for predicting myocardial infarction using multiple data modalities.
Proposed GFM-MIP integrates ECG time-series, ECG images, and laboratory test results.
Utilized a Graphormer encoder to model dependencies in ECG signals.
Employed a Vision Transformer for morphological pattern extraction from images.
Incorporated Feature-wise Linear Modulation to tailor the model to patient-specific features.
Applied a Transformer-based fusion module to analyze interactions between multiple data types.
GFM-MIP consistently outperformed existing baselines in performance metrics.
Validation studies showed the importance of each data modality in the predictions.
Achieved robust results on a real-world clinical dataset and multiple public benchmarks.

Structured PICO

Does the GFM-MIP multimodal fusion framework improve myocardial infarction prediction compared to state-of-the-art baselines?

Population

Datasets containing 12-lead ECG time-series signals, ECG images, and laboratory test results (one real-world clinical dataset and three public benchmarks)

Intervention

GFM-MIP (Graph-informed and FiLM-enhanced Multimodal Fusion framework)

Comparator

State-of-the-art baseline models

Outcome

Myocardial infarction prediction performancesurrogate

A novel multimodal fusion framework integrating ECG signals, images, and lab tests improves the prediction of myocardial infarction over existing single-modality methods.

Main Result

Absolute Event Rate: 0% vs 0%

Abstract

Accurate and timely diagnosis of cardiovascular diseases, particularly myocardial infarction (MI), remains a critical clinical challenge. Existing electrocardiogram (ECG) analysis methods often rely solely on a single data modality, such as raw signals or waveform images, which limits their ability to capture the broader physiological context. To address this limitation, we propose GFM-MIP, a Graph-informed and FiLM-enhanced Multimodal Fusion framework for myocardial infarction prediction. GFM-MIP integrates 12-lead ECG time-series signals, ECG images, and laboratory test results through a unified architecture. Specifically, it employs a Graphormer encoder to model inter-lead dependencies in ECG signals and a Vision Transformer to extract morphological patterns from ECG images, both modulated by patient-specific laboratory features using Feature-wise Linear Modulation (FiLM). A Transformer-based fusion module captures cross-modal interactions, while a contrastive learning objective encourages alignment between signal and image modalities. Experimental results on a real-world clinical dataset and three public benchmarks demonstrate that GFM-MIP consistently outperforms state-of-the-art baselines across multiple evaluation metrics. Ablation studies further validate the contribution of each modality and architectural component. The proposed framework offers a clinically meaningful and scalable solution for robust, multimodal cardiovascular diagnosis.

KI fragen

Bookmark

View Full Paper