DERI: Cross-Modal ECG Representation Learning with Deep ECG-Report Interaction

Authors: Jian Chen, Xiaoru Dong, Wei Wang, Shaorui Zhou, Lequan Yu, Xiping Hu

IJCAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Comprehensive experiments with various settings are conducted on various datasets to show the superior performance of our DERI.
Researcher Affiliation Academia 1Artificial Intelligence Research Institute, Shenzhen MSU-BIT University 2School of Intelligent System Engineering, Sun Yat-sen University 3School of Computing and Data Science, The University of Hong Kong 4School of Medical Technology, Beijing Institute of Technology EMAIL, xrdong,EMAIL, EMAIL, EMAIL, EMAIL
Pseudocode No The paper describes its methodology in prose and mathematical equations in Section 3 (Methodology) but does not include any clearly labeled pseudocode or algorithm blocks.
Open Source Code Yes Our code is released on https://github.com/cccccj-03/DERI.
Open Datasets Yes MIMIC-ECG. The pre-training process of our proposed DERI model is conducted on the MIMIC-ECG dataset [Gow et al., 2023]... PTBXL [Wagner et al., 2020] contains 21,837 12-lead ECG signals... CPSC2018 [Liu et al., 2018] contains 6,877 12-lead ECG recordings... Chapman-Shaoxing-Ningbo (CSN) [Zheng et al., 2022] contains 45,152 12-lead ECG recordings...
Dataset Splits Yes For linear probing, we add a new linear classifier and freeze all other parameters in our DERI. We adopt three different settings, which utilize 1%, 10%, and 100% of the training data.
Hardware Specification Yes We conduct all the pre-trained experiments on 4 NVIDIA GeForce RTX 4090 GPUs with a batch size of 512.
Software Dependencies No For the encoders used for ECG signals and reports, we adopt a randomly initialized 1D-ResNet18 and the Med-CPT [Jin et al., 2023], respectively... After we obtain the final representation of ECG, we adopt GPT-2 as the text decoder to construct an encoder-decoder structure since Distil GPT2 [Sanh, 2019] has shown its great performance on report generation [Wang et al., 2024].
Experiment Setup Yes We use the AdamW optimizer with a learning rate of 1e-3 and a weight decay of 1e-8. The epoch for pre-training is set as 50 with a cosine annealing scheduler to adjust the learning rate. We conduct all the pre-trained experiments on 4 NVIDIA GeForce RTX 4090 GPUs with a batch size of 512.