What question did this study set out to answer?

The aim is to develop an ECG encoding method that overcomes limitations of existing model-specific encoders and improves interpretability in LLMs.

January 10, 2026Computers in Biology and Medicine0 citations

View Full Paper

ECG-aBcDe: Overcoming model dependence, encoding ECG into a universal language for any large language model

YXYong XiaHarbin Institute of Technology JLJingxuan LiHarbin Institute of Technology YSYunqi SunQingdao University

Key Result

The ECG-aBcDe method outperforms existing ECG encoding methods, achieving a Bleu-4 score of 42.58, enhancing LLM interpretability and temporal modeling in ECG analysis.

Key Points

The aim is to develop an ECG encoding method that overcomes limitations of existing model-specific encoders and improves interpretability in LLMs.
Introduced ECG-aBcDe for universal ECG encoding applicable to any LLM.
Constructed a hybrid dataset of ECG language and natural language for training models.
Achieved bidirectional convertibility to enhance interpretability with attention heatmaps.
Evaluated model performance using Rouge-L, Meteor, and Bleu-4 metrics.
ECG-aBcDe achieved competitive Rouge-L and Meteor scores while significantly outperforming existing methods on Bleu-4.
Bleu-4 score reached 42.58, demonstrating effective temporal modeling capabilities.
The method enhances interpretability of ECG signals for clinical applications.

Structured PICO

Population

ECG datasets (specifics not provided in abstract)

Intervention

ECG-aBcDe, a novel ECG encoding method that transforms ECG signals into a universal ECG language for Large Language Models (LLMs)

Comparator

Existing ECG encoding methods

Outcome

Bleu-4, Rouge-L, and Meteor scoressurrogate

ECG-aBcDe is a novel encoding method that translates ECG signals into a universal language for LLMs, improving interpretability and temporal modeling without requiring architectural modifications.

Main Result

Absolute Event Rate: 0% vs 0%

Abstract

Large Language Models (LLMs) hold significant promise for electrocardiogram (ECG) analysis, yet challenges remain regarding transferability, time-scale information learning, and interpretability. Current methods suffer from model-specific ECG encoders, hindering transfer across LLMs. Furthermore, LLMs struggle to capture crucial time-scale information inherent in ECGs due to Transformer limitations. And their black-box nature limits clinical adoption. To address these limitations, we introduce ECG-aBcDe, a novel ECG encoding method that transforms ECG signals into a universal ECG language readily interpretable by any LLM. By constructing a hybrid dataset of ECG language and natural language, ECG-aBcDe enables direct fine-tuning of pre-trained LLMs without architectural modifications, achieving "construct once, use anywhere" capability. Moreover, the bidirectional convertibility between ECG and ECG language of ECG-aBcDe allows for extracting attention heatmaps from ECG signals, significantly enhancing interpretability. Finally, ECG-aBcDe explicitly represents time-scale information, mitigating Transformer limitations. This work presents a new paradigm for integrating ECG analysis with LLMs. Compared with existing methods, our approach achieves competitive Rouge-L and Meteor scores and significantly outperforms them on Bleu-4, reaching 42.58 and 30.76, which demonstrates the effectiveness and feasibility of the proposed paradigm. The proposed ECG-aBcDe method enhances the temporal modeling capability and interpretability of LLMs in ECG analysis, providing a robust foundation for future clinical decision support systems.

问 AI

Bookmark

View Full Paper

Cite This Study

Xia et al. (Fri,) reported a other. The ECG-aBcDe method outperforms existing ECG encoding methods, achieving a Bleu-4 score of 42.58, enhancing LLM interpretability and temporal modeling in ECG analysis.

synapsesocial.com/papers/696321df91e05aa366cb828f https://doi.org/https://doi.org/10.1016/j.compbiomed.2025.111439

问 AI

Bookmark

View Full Paper