Q-Heart: ECG Question Answering via Knowledge-Informed Multimodal LLMs

Aaqib Saeed; Dong Ma; Hung Manh Pham; Jialu Tang

REVIEW 3 major objections 5 minor 42 references

Q-Heart: ECG Question Answering via Knowledge-Informed Multimodal LLMs

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Q-Heart claims that an ECG question-answering system can reach state-of-the-art exact-match accuracy by combining a lead-aware mapping layer, retrieval of historical reports, and dynamic instruction tuning of a compact language model.

desk verdict A solid SOTA engineering result for ECG-QA, but the retrieval gain isn't isolated from a mere-text effect and the evaluation is single-run. read the letter →

arxiv 2505.06296 v1 pith:GBIJ7SIY submitted 2025-05-07 eess.SP

Hung Manh Pham , Jialu Tang , Aaqib Saeed , Dong Ma This is my paper · ORCID

classification eess.SP

keywords ECGquestionansweringmultimodallargelanguagemodelsrepresentationlearningretrieval-augmentedgenerationinstructiontuninglead-awaremapping12-leadexactmatchaccuracy

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Q-Heart is a multimodal system that answers clinical questions about raw 12-lead ECG recordings, and the paper's claim is that its design choices—a strong ECG encoder, a lead-aware transformer mapper, and retrieval of historical reports during instruction tuning—are what push accuracy past prior methods. On the benchmark task, the system reports a 0.614 average exact-match accuracy on the full test set and 0.628 on a 10% subset, roughly four percentage points above the next-best published result. The improvement matters because ECG question answering is still in its early stages: most existing systems either classify with weak encoders, rely on patient-specific metadata, or use large proprietary models with undisclosed ECG processing. Q-Heart instead shows that a compact, instruction-tuned language model grounded in retrieved clinical reports can handle the verify and choose question types with high accuracy, while open-ended query questions remain a recognized weakness.

What carries the argument

The load-bearing object is the ET-Mapper, a lightweight transformer that converts the ECG encoder's pooled embedding into a sequence of twelve prefix embeddings, one per standard lead, sized to match the language model's token space. It is made lead-aware by a skip connection that carries the encoder's per-lead positional embedding through a convolution and linear projection and adds it to the mapped prefix. The other central mechanism is retrieval-augmented dynamic prompting: historical clinical reports are indexed by the same encoder, the query ECG retrieves its top three by cosine similarity, and the prompt randomly includes one of them while the answer options are shuffled; this trains the model to ground answers in evidence rather than memorize option positions.

What would settle it

Give the model the same training and prompts but replace the retrieved top-three reports with three randomly selected historical reports; if exact-match accuracy on the 10% test subset stays at about 0.628, the retrieval step's clinical relevance is not what drives the reported gain.

Watch

Extended reading notes

Core claim

At the core, the paper argues that ECG question answering can be treated as a generative multimodal alignment problem rather than a classification problem. A 12-lead ECG is encoded by C-MELTv2, a transformer-based encoder pretrained from scratch with aggressive random lead masking, into a single 768-dimensional vector. The ET-Mapper, a two-layer, four-head transformer with a skip-connected positional embedding from the encoder, expands that vector into twelve LLM-compatible prefix tokens, one per lead, so the language model receives both global cardiac context and lead-specific spatial cues. In parallel, the same encoder embedding retrieves the top three historical ECG reports from a large indexed database by cosine similarity, and these reports are inserted into a dynamically constructed prompt, with answer options shuffled and one of the three reports randomly chosen at each training step. The resulting instruction-tuned model reaches 0.909 verify and 0.603 choose exact-match accuracy on the full test set, and the ablations attribute the largest single drop, up to four points in exact-match, to removing the lead-positional skip connection. The paper's own numbers also show query-type questions, which often require measuring numeric intervals, remain below classification-based baselines.

Load-bearing premise

The pipeline assumes that a single pooled 768-dimensional embedding of a 12-lead ECG still preserves enough per-lead detail for cosine-similarity retrieval to find clinically relevant historical reports and for the lead-aware mapper to convey which lead a question is about.

Editorial extensions

If this is right

If the reported results hold, verify-style questions—'Does this ECG show X in lead I?'—are largely solvable, with accuracy near 0.91, making automated triage or screening for clearly named findings plausible.
The ablation evidence implies that per-lead positional information is not a minor detail: removing it costs up to four exact-match points, so any ECG-QA architecture that pools all leads before entering the language model is likely leaving accuracy on the table.
Retrieval of historical reports contributes about two to three points on average, suggesting that grounding an LLM in previously written clinical reports can compensate for the small size of an instruction-tuned backbone.
A compact 1B-parameter language model tuned with low-rank adapters is sufficient to beat much larger frozen medical models on this benchmark, which lowers the compute barrier for clinical ECG-QA systems.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

A natural next test, given the retrieval design, is to check whether the retrieved reports themselves are medically plausible for the input ECG; if they are, the pipeline doubles as an interpretability tool that shows clinicians which historical cases the answer was based on.
Because questions about numeric intervals still fail, injecting standard hand-computed ECG features—heart rate, PR/QRS/QT intervals, noise flags—into the prompt is the most direct way to close the query gap; the paper lists this as future work.
The design's dependence on twelve lead-aligned prefix tokens suggests it may not transfer unchanged to single- or few-lead wearable ECGs; testing on reduced-lead inputs would reveal whether the lead-aware skip connection can adapt.
The option-shuffling component shows only a small gain, which suggests positional bias is not the main error source; ablating retrieval by substituting random reports (rather than removing reports) would isolate whether clinical similarity or extra text causes the improvement.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Here's my read on Q-Heart. The short version: this is a reasonably solid engineering contribution that reports a genuine SOTA on the ECG-QA benchmark, but the evaluation has two soft spots that need airing before I'd fully trust the claims—the retrieval component is not isolated from a mere-text effect, and all results are single-run with no error bars.

What's actually new: they take their own C-MELT encoder, extend it (12 transformer layers, heavier lead masking), and validate it zero-shot on PTB-XL and CODE-15, where it outperforms the original C-MELT and MERL. That validation is real and independent of the QA task. The ET-Mapper with the lead-aware positional skip connection is a sensible design choice, and the dynamic prompt shuffling plus random selection among retrieved reports is a reasonable way to regularize instruction tuning. The reported 0.614 average EM-Acc on the full test set and 0.628 on the 10% subset beat the previous best by around 4 points, which is honest progress in a specialized subfield.

Where it gets softer. Table 4's ablation removes the retrieved report entirely but never replaces it with an irrelevant one. So the 2.3-point gain from retrieval could just be the presence of any extra text in the prompt, not clinical relevance. That's a real gap, and it's easy to fix—a random-report control, or at least reporting retrieval hit rates, would settle it. There's also a worry that for verify/choose questions, the answer string could appear verbatim in retrieved MIMIC-IV reports, giving the model a shortcut that non-retrieval baselines don't have. The paper doesn't report how often that happens, so the "knowledge-informed" mechanism is not yet pinned down. I don't think this invalidates the SOTA claim, because the full system includes retrieval, but it does weaken the interpretation.

The other soft spots are more minor: single-run numbers without confidence intervals, no code release, and the query-type accuracy (0.329) sitting below several classification baselines. Their defense—generative vs. classification framing—is partially fair, but it's a caveat worth noting. The lead-aware skip connection ablation shows it helps, but that could partly be an effect of extra parameters, so I'd like to see it tested against a similarly-sized alternative.

Overall: the central result likely holds, but the paper's scientific claims are a bit ahead of its evidence. It's for people working on ECG-language models and medical multimodal LLMs. I'd send this to review, but I'd ask the authors for reproducibility details and that retrieval control experiment before accepting.

Referee Report

3 major / 5 minor

Summary. The manuscript proposes Q-Heart, a multimodal LLM framework for ECG-based question answering. The system combines a 12-layer ECG encoder adapted from C-MELT (trained from scratch on MIMIC-IV ECG), a transformer-based ET-Mapper that converts the ECG embedding into 12 lead-conditioned prefix tokens, retrieval of top-3 historical reports from a FAISS index over MIMIC-IV ECG-report pairs, and instruction tuning of LLaMA-3.2-1B-Instruct with LoRA and dynamic prompting. On the ECG-QA benchmark, Q-Heart reports average exact-match accuracy of 0.614 on the full test set and 0.628 on a 10% subset, exceeding the previous best averages by about 3-4 points. Per-type results are strongest for verify questions (0.909 full / 0.913 10%) and weakest for query questions (0.329 full / 0.345 10%). Ablations in Table 4 attribute gains to dynamic prompting, retrieval, the ET-Mapper, lead positional encoding, and fine-tuning of both the LLM and the ECG encoder.

Significance. If the reported results are reproducible and the retrieval mechanism is proven to be knowledge-grounded rather than a generic prompt effect or answer leak, Q-Heart would be a solid contribution: it is a relatively lightweight 1B-parameter ECG-QA system that outperforms the previous best average exact-match accuracy on a public benchmark, and it validates its ECG encoder zero-shot on PTB-XL and CODE-15, including a comparison with human expert performance. The ablation matrix is a genuine strength, as each component is removed in a controlled within-model comparison. The central significance hinges on whether the retrieval-augmentation gain reflects a clinically organized embedding space and actual retrieval-based reasoning; the current manuscript does not yet establish this conclusively.

major comments (3)

[Table 4; Sections 3.1 and 3.3] The retrieval-augmentation claim is not isolated by a proper control condition. The 'W/o Retrieval Report' ablation removes the report from the prompt entirely, but it does not replace it with an irrelevant or randomly selected report from the same FAISS index; the observed 2.3-point EM-Acc drop (0.628 vs. 0.605) could therefore be caused simply by the presence of additional text in the prompt, not by clinical similarity between the query ECG and the retrieved historical case. Moreover, because retrieved MIMIC-IV reports are inserted verbatim, the ground-truth answer for verify and choose questions (e.g., 'noise in lead I', 'yes', or a specific lead name) may appear directly in the retrieved text; if this happens frequently, non-retrieval baselines are not operating on the same information channel, and the SOTA margin is not evidence of ECG-grounded reasoning. The paper should report retrieval-quality statistics (e.g., how often the answer string appears in the top-3 retrieved reports, per question type) and add a random-report control condition.
[Table 3, Section 4.4.2] The per-type results contradict the text's characterization of query performance. On the full test set, Q-Heart obtains 0.329 query EM-Acc, which is 8.1 points below M3AE (0.410) and 7.5 points below MedViLL (0.404); on the 10% subset it obtains 0.345 versus 0.400 for M3AE. This is not a slight underperformance, and the claimed SOTA average is driven by a very large verify margin (0.909 vs. 0.746 for M3AE). The paper should either compare against generative baselines with matched output spaces or substantially qualify the state-of-the-art claim; the current framing understates a real weakness of the method on open-ended query questions.
[Tables 3-4, Figures 5-6] All quantitative claims rest on single-run evaluations without error bars, confidence intervals, or significance tests. This matters because the central SOTA margin is about 3 points on EM-Acc, and several ablation differences in Table 4 are 1-2 points, which could easily be within run-to-run variance for instruction-tuned models. Please report multiple seeds or bootstrap confidence intervals for the main results and ablations, and state the number of runs used.

minor comments (5)

[Table 2 caption] The caption phrase 'lead care' should likely be 'lead questions'; the misspellings 'archieves' (Section 4.4.1) and 'hybridly combines' (Section 2.1) should also be corrected.
[Table 3] The Med-Gemini rows report only average values without per-type breakdowns or a description of the fine-tuning protocol; without this information, the comparison against fine-tuned Med-Gemini cannot be fully assessed.
[Section 3.2] The derivation of the lead positional representation pe from the ECG encoder's positional embedding layer is not specified; please state the exact tensor shape and the convolutional/linear transformations used so that the ET-Mapper can be reimplemented.
[Section 4.2] The 10% subset of ECG-QA is not defined with a sampling seed or splitting procedure; please specify how it was selected and whether it is identical to the setup used in the original ECG-QA benchmark paper.
[General] The manuscript does not state whether code or model weights will be released; for a systems paper with many training details, a code release statement would materially aid reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: Q-HEART's SOTA claim is grounded in an external benchmark (ECG-QA) and component-wise ablations; the retrieval-control gap is an experimental-validity concern, not a by-construction reduction.

full rationale

Q-HEART's central claim (0.614 average EM-Acc on the full ECG-QA test set, Table 3) is measured against a held-out test split of an external benchmark, so the reported accuracy is not a fitted input. The ablation study (Table 4) compares the full system with real ablations (w/o DP, w/o Retrieval Report, w/o ET-Mapper, w/o Pos-Encoder, frozen LLM, frozen ECG encoder); each removal changes a component and the metric moves, which is evidence that the components are doing measurable work. The ECG encoder is adapted from the authors' own C-MELT [25], a self-citation, but C-MELTv2 is trained from scratch ('we trained entirely from scratch without relying on any pretrained weights') and validated zero-shot on independent PTB-XL and CODE-15 datasets (Figures 5-6), so the architecture choice is supported by externally checkable results, not by an unverified self-citation chain. The reviewer concern that the W/o Retrieval Report ablation lacks an irrelevant-report control condition, and that retrieved MIMIC-IV reports might contain answer-relevant text, is a legitimate threat to the interpretation of the retrieval gain; however, it is a question of experimental control and possible information leakage, not circularity, because no equation defines the reported EM-Acc in terms of the retrieved report, and no fitted parameter is renamed as a prediction. No uniqueness theorem from the authors' prior work is invoked, and no known result is merely renamed. On balance, the derivation chain is self-contained against external benchmarks, so the circularity score is low.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on standard supervised training and evaluation, so the ledger is dominated by hyperparameters and domain assumptions rather than free physical constants. There are no invented physical entities. The nontrivial assumptions are that the pretrained embedding space supports clinically meaningful similarity retrieval, that the pooled ECG embedding retains lead-specific information for the skip connection, and that the benchmark was not leaked into pretraining or retrieval. These assumptions are plausible but not directly verified in the paper.

free parameters (5)

ET-Mapper prefix length c = 12
The mapping projects the global ECG embedding into 12 prefix tokens, one per lead. Chosen by design to mirror the 12 leads; no sensitivity analysis is reported.
Lead masking probability p = 0.65
Each lead is masked with probability 0.65 during ECG encoder pretraining. Motivated by the observation that about 60% of ECG-QA questions mention a specific lead; no sensitivity study is given.
Number of retrieved reports top-k = 3
The retriever returns the top-3 similar reports, and one is randomly sampled per training step. The value 3 is stated without reported tuning.
ET-Mapper transformer size (L, H) = L=2, H=4
The mapping transformer uses 2 layers and 4 attention heads. These values are presented as fixed design choices without ablation.
LoRA rank r, alpha, dropout = r=8, alpha=32, dropout=0.1
Standard LoRA configuration used for LLM fine-tuning; no sensitivity analysis is reported.

assumptions (4)

domain assumption Cosine similarity in the pretrained C-MELTv2 embedding space retrieves clinically relevant historical reports.
Section 3.1 assumes this for FAISS index construction and query-time retrieval, but no retrieval quality analysis, precision, or failure cases are provided.
domain assumption The pooled ECG embedding ze retains enough per-lead information for the positional skip connection pe to be meaningful.
Section 3.2 extracts pe from the encoder's positional embedding layer and adds it to the mapped prefix. This only helps if ze has not fully discarded lead-level structure during global pooling.
domain assumption The ECG-QA test set and its labels were not used to build the FAISS retrieval index or to select hyperparameters beyond the validation split.
Section 4.1 and 4.2 say model selection uses validation loss and the retrieval index is built from MIMIC-IV ECG, a different cohort, so leakage is unlikely but not audited.
domain assumption ECG-QA questions are answerable from the 12-lead signal plus retrieved reports, without patient metadata.
The method deliberately avoids patient-specific data, so the benchmark must not require such data for correct answers. The paper provides no explicit check of this property.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Q-Heart: ECG Question Answering via Knowledge-Informed Multimodal LLMs." pith.science (2026). https://pith.science/paper/GBIJ7SIY

@misc{pith2026250506296,
  author       = {Pith},
  title        = {Pith review of: Q-Heart: ECG Question Answering via Knowledge-Informed Multimodal LLMs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GBIJ7SIY}},
  note         = {Machine review of arXiv:2505.06296}
}

read the original abstract

Electrocardiography (ECG) offers critical cardiovascular insights, such as identifying arrhythmias and myocardial ischemia, but enabling automated systems to answer complex clinical questions directly from ECG signals (ECG-QA) remains a significant challenge. Current approaches often lack robust multimodal reasoning capabilities or rely on generic architectures ill-suited for the nuances of physiological signals. We introduce Q-Heart, a novel multimodal framework designed to bridge this gap. Q-Heart leverages a powerful, adapted ECG encoder and integrates its representations with textual information via a specialized ECG-aware transformer-based mapping layer. Furthermore, Q-Heart leverages dynamic prompting and retrieval of relevant historical clinical reports to guide tuning the language model toward knowledge-aware ECG reasoning. Extensive evaluations on the benchmark ECG-QA dataset show Q-Heart achieves state-of-the-art performance, outperforming existing methods by a 4% improvement in exact match accuracy. Our work demonstrates the effectiveness of combining domain-specific architectural adaptations with knowledge-augmented LLM instruction tuning for complex physiological ECG analysis, paving the way for more capable and potentially interpretable clinical patient care systems.

Figures

Figures reproduced from arXiv: 2505.06296 by the authors.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 26 canonical work pages

[1]

Automated Medical Report Generation for ECG Data: Bridging Medical Text and Signal Processing with Deep Learning

A. Bleich, A. Linnemann, B. H. Diem, and T. O. Conrad. Automated medical report generation for ecg data: Bridging medical text and signal processing with deep learning. arXiv preprint arXiv:2412.04067, 2024

work page Pith review arXiv 2024
[2]

Z. Chen, Y . Du, J. Hu, Y . Liu, G. Li, X. Wan, and T.-H. Chang. Multi-modal masked autoencoders for medical vision-and-language pre- training. In International Conference on Medical Image Computing and Computer-Assisted Intervention, pages 679–689. Springer, 2022

work page 2022
[3]

Denkowski and A

M. Denkowski and A. Lavie. Meteor 1.3: Automatic metric for reli- able optimization and evaluation of machine translation systems. In C. Callison-Burch, P. Koehn, C. Monz, and O. F. Zaidan, editors,Pro- ceedings of the Sixth Workshop on Statistical Machine Translation, pages 85–91, Edinburgh, Scotland, July 2011. Association for Computational Linguistics....

work page 2011
[4]

Douze, A

M. Douze, A. Guzhva, C. Deng, J. Johnson, G. Szilvasy, P.-E. Mazaré, M. Lomeli, L. Hosseini, and H. Jégou. The faiss library. arXiv preprint arXiv:2401.08281, 2024

arXiv 2024
[5]

Gopal, R

B. Gopal, R. Han, G. Raghupathi, A. Ng, G. Tison, and P. Rajpurkar. 3kg: Contrastive learning of 12-lead electrocardiograms using physiologically- inspired augmentations. In Machine learning for health, pages 156–167. PMLR, 2021

work page 2021
[6]

B. Gow, T. Pollard, L. A. Nathanson, A. Johnson, B. Moody, C. Fer- nandes, N. Greenbaum, S. Berkowitz, D. Moukheiber, P. Eslami, et al. Mimic-iv-ecg-diagnostic electrocardiogram matched subset. Type: dataset, 2023

work page 2023
[7]

H. Han, S. Park, S. Min, H.-S. Choi, E. Kim, H. Kim, S. Park, J. Kim, J. Park, J. An, et al. Towards high generalization performance on elec- trocardiogram classification. In 2021 Computing in Cardiology (CinC), volume 48, pages 1–4. IEEE, 2021

work page 2021
[8]

E. J. Hu, Y . Shen, P. Wallis, Z. Allen-Zhu, Y . Li, S. Wang, L. Wang, and W. Chen. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685, 2021

arXiv 2021

Show all 42 references

[9]

R. Hu, J. Chen, and L. Zhou. Spatiotemporal self-supervised representa- tion learning from multi-lead ecg signals. Biomedical Signal Processing and Control, 84:104772, 2023

2023
[10]

X. Hu, L. Gu, K. Kobayashi, L. Liu, M. Zhang, T. Harada, R. M. Sum- mers, and Y . Zhu. Interpretable medical image visual question answering via multi-modal relationship graph learning. Medical Image Analysis, 97:103279, 2024

2024
[11]

E. Jing, H. Zhang, Z. Li, Y . Liu, Z. Ji, and I. Ganchev. Ecg heartbeat classification based on an improved resnet-18 model. Computational and Mathematical Methods in Medicine, 2021(1):6649970, 2021

2021
[12]

Kiyasseh, T

D. Kiyasseh, T. Zhu, and D. A. Clifton. Clocs: Contrastive learning of cardiac signals across space, time, and patients. In International Conference on Machine Learning, pages 5606–5615. PMLR, 2021

2021
[13]

S. K. Lalam, H. K. Kunderu, S. Ghosh, H. Kumar, S. Awasthi, A. Prasad, F. Lopez-Jimenez, Z. I. Attia, S. Asirvatham, P. Friedman, et al. Ecg representation learning with multi-modal ehr data. Transactions on Machine Learning Research, 2023

2023
[14]

C.-Y . Lin. Rouge: A package for automatic evaluation of summaries. In Text summarization branches out, pages 74–81, 2004

2004
[15]

C. Liu, Z. Wan, C. Ouyang, A. Shah, W. Bai, and R. Arcucci. Zero- shot ecg classification with multimodal learning and test-time clinical knowledge enhancement. In Forty-first International Conference on Machine Learning
[16]

C. Liu, Z. Wan, S. Cheng, M. Zhang, and R. Arcucci. Etp: Learning transferable ecg representations via ecg-text pre-training. In ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 8230–8234. IEEE, 2024

2024
[17]

Y . Luo, J. Zhang, S. Fan, K. Yang, Y . Wu, M. Qiao, and Z. Nie. Biomedgpt: Open multimodal generative pre-trained transformer for biomedicine. arXiv preprint arXiv:2308.09442, 2023

2023 arXiv
[18]

Llama 3.2 1b instruct, 2024

Meta AI. Llama 3.2 1b instruct, 2024. URL https://huggingface.co/ meta-llama/Llama-3.2-1B-Instruct

2024
[19]

J. H. Moon, H. Lee, W. Shin, Y .-H. Kim, and E. Choi. Multi-modal under- standing and generation for medical images and text via vision-language pre-training. IEEE Journal of Biomedical and Health Informatics, 26 (12):6070–6080, 2022

2022
[20]

Y . Na, M. Park, Y . Tae, and S. Joo. Guiding masked representation learning to capture spatio-temporal relationship of electrocardiogram. arXiv preprint arXiv:2402.09450, 2024

2024 arXiv
[21]

J. Oh, H. Chung, J.-m. Kwon, D.-g. Hong, and E. Choi. Lead-agnostic self-supervised learning for local and global representations of electro- cardiogram. In Conference on Health, Inference, and Learning, pages 338–353. PMLR, 2022

2022
[22]

J. Oh, G. Lee, S. Bae, J.-m. Kwon, and E. Choi. Ecg-qa: A comprehensive question answering dataset combined with electrocardiogram. In A. Oh, T. Neumann, A. Globerson, K. Saenko, M. Hardt, and S. Levine, editors, Advances in Neural Information Processing Systems, volume 36, pa...

2023
[23]

J. W. O’Sullivan, A. Palepu, K. Saab, W.-H. Weng, Y . Cheng, E. Chu, Y . Desai, A. Elezaby, D. S. Kim, R. Lan, et al. Towards democratization of subspeciality medical expertise. arXiv preprint arXiv:2410.03741, 2024

2024 arXiv
[24]

Papineni, S

K. Papineni, S. Roukos, T. Ward, and W.-J. Zhu. Bleu: a method for automatic evaluation of machine translation. In Proceedings of the 40th annual meeting of the Association for Computational Linguistics, pages 311–318, 2002

2002
[25]

M. Pham, A. Saeed, and D. Ma. C-melt: Contrastive enhanced masked auto-encoders for ecg-language pre-training. arXiv preprint arXiv:2410.02131, 2024

2024 arXiv
[26]

J. Qiu, J. Zhu, S. Liu, W. Han, J. Zhang, C. Duan, M. A. Rosenberg, E. Liu, D. Weber, and D. Zhao. Automated cardiovascular record retrieval by multimodal learning between electrocardiogram and clinical report. In Machine Learning for Health (ML4H), pages 480–497. PMLR, 2023

2023
[27]

A. H. Ribeiro, M. H. Ribeiro, G. M. Paixão, D. M. Oliveira, P. R. Gomes, J. A. Canazart, M. P. Ferreira, C. R. Andersson, P. W. Mac- farlane, W. Meira Jr, et al. Automatic diagnosis of the 12-lead ecg using a deep neural network. Nature communications, 11(1):1760, 2020

2020
[28]

K. Saab, T. Tu, W.-H. Weng, R. Tanno, D. Stutz, E. Wulczyn, F. Zhang, T. Strother, C. Park, E. Vedadi, et al. Capabilities of gemini models in medicine. arXiv preprint arXiv:2404.18416, 2024

2024 arXiv
[29]

J. Tang, T. Xia, Y . Lu, C. Mascolo, and A. Saeed. Electrocardiogram- language model for few-shot question answering with meta learning. arXiv preprint arXiv:2410.14464, 2024

2024 arXiv
[30]

J. Tang, T. Xia, Y . Lu, C. Mascolo, and A. Saeed. Electrocardiogram report generation and question answering via retrieval-augmented self- supervised modeling. In ICASSP 2025-2025 IEEE International Confer- ence on Acoustics, Speech and Signal Processing (ICASSP), pages 1–5. ...

2025
[31]

Wagner, N

P. Wagner, N. Strodthoff, R.-D. Bousseljot, D. Kreiseler, F. I. Lunze, W. Samek, and T. Schaeffter. Ptb-xl, a large publicly available electro- cardiography dataset. Scientific data, 7(1):1–15, 2020

2020
[32]

Z. Wan, C. Liu, X. Wang, C. Tao, H. Shen, Z. Peng, J. Fu, R. Arcucci, H. Yao, and M. Zhang. Meit: Multi-modal electrocardiogram instruction tuning on large language models for report generation. arXiv preprint arXiv:2403.04945, 2024

2024 arXiv
[33]

X. Wang, J. Kang, P. Han, Y . Zhao, Q. Liu, L. He, L. Zhang, L. Dai, Y . Wang, and J. Tao. Ecg-expert-qa: A benchmark for evaluating med- ical large language models in heart disease diagnosis. arXiv preprint arXiv:2502.17475, 2025

2025 arXiv
[34]

Q. Yan, J. Duan, and J. Wang. Multi-modal concept alignment pre- training for generative medical visual question answering. In Findings of the Association for Computational Linguistics ACL 2024, pages 5378– 5389, 2024

2024
[35]

K. Yang, M. Hong, J. Zhang, Y . Luo, Y . Su, O. Zhang, X. Yu, J. Zhou, L. Yang, M. Qian, et al. Ecg-lm: Understanding electrocardiogram with large language model. Health Data Science, 2024

2024
[36]

H. Yu, P. Guo, and A. Sano. Zero-shot ecg diagnosis with large language models and retrieval-augmented generation. In Machine Learning for Health (ML4H), pages 650–663. PMLR, 2023

2023
[37]

H. Yu, P. Guo, and A. Sano. Ecg semantic integrator (esi): A founda- tion ecg model pretrained with llm-enhanced cardiological text. arXiv preprint arXiv:2405.19366, 2024

2024 arXiv
[38]

X. Zhai, B. Mustafa, A. Kolesnikov, and L. Beyer. Sigmoid loss for lan- guage image pre-training. In Proceedings of the IEEE/CVF International Conference on Computer Vision, pages 11975–11986, 2023

2023
[39]

Zhang, V

T. Zhang, V . Kishore, F. Wu, K. Q. Weinberger, and Y . Artzi. Bertscore: Evaluating text generation with bert. arXiv preprint arXiv:1904.09675, 2019

1904 arXiv
[40]

Zhang, L

W. Zhang, L. Yang, S. Geng, and S. Hong. Self-supervised time se- ries representation learning via cross reconstruction transformer. IEEE Transactions on Neural Networks and Learning Systems, 2023

2023
[41]

Y . Zhao, T. Zhang, X. Wang, P. Han, T. Chen, L. Huang, Y . Jin, and J. Kang. Ecg-chat: A large ecg-language model for cardiac disease diagnosis. arXiv preprint arXiv:2408.08849, 2024

2024 arXiv
[42]

Zhong, O

X. Zhong, O. Gong, W. Huang, L. Li, and H. Xia. Squeeze-and-excitation wide residual networks in image classification. In 2019 IEEE Interna- tional Conference on Image Processing (ICIP), pages 395–399. IEEE, 2019

2019

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [1]

Automated Medical Report Generation for ECG Data: Bridging Medical Text and Signal Processing with Deep Learning

A. Bleich, A. Linnemann, B. H. Diem, and T. O. Conrad. Automated medical report generation for ecg data: Bridging medical text and signal processing with deep learning. arXiv preprint arXiv:2412.04067, 2024

work page Pith review arXiv 2024

[2] [2]

Z. Chen, Y . Du, J. Hu, Y . Liu, G. Li, X. Wan, and T.-H. Chang. Multi-modal masked autoencoders for medical vision-and-language pre- training. In International Conference on Medical Image Computing and Computer-Assisted Intervention, pages 679–689. Springer, 2022

work page 2022

[3] [3]

Denkowski and A

M. Denkowski and A. Lavie. Meteor 1.3: Automatic metric for reli- able optimization and evaluation of machine translation systems. In C. Callison-Burch, P. Koehn, C. Monz, and O. F. Zaidan, editors,Pro- ceedings of the Sixth Workshop on Statistical Machine Translation, pages 85–91, Edinburgh, Scotland, July 2011. Association for Computational Linguistics....

work page 2011

[4] [4]

Douze, A

M. Douze, A. Guzhva, C. Deng, J. Johnson, G. Szilvasy, P.-E. Mazaré, M. Lomeli, L. Hosseini, and H. Jégou. The faiss library. arXiv preprint arXiv:2401.08281, 2024

arXiv 2024

[5] [5]

Gopal, R

B. Gopal, R. Han, G. Raghupathi, A. Ng, G. Tison, and P. Rajpurkar. 3kg: Contrastive learning of 12-lead electrocardiograms using physiologically- inspired augmentations. In Machine learning for health, pages 156–167. PMLR, 2021

work page 2021

[6] [6]

B. Gow, T. Pollard, L. A. Nathanson, A. Johnson, B. Moody, C. Fer- nandes, N. Greenbaum, S. Berkowitz, D. Moukheiber, P. Eslami, et al. Mimic-iv-ecg-diagnostic electrocardiogram matched subset. Type: dataset, 2023

work page 2023

[7] [7]

H. Han, S. Park, S. Min, H.-S. Choi, E. Kim, H. Kim, S. Park, J. Kim, J. Park, J. An, et al. Towards high generalization performance on elec- trocardiogram classification. In 2021 Computing in Cardiology (CinC), volume 48, pages 1–4. IEEE, 2021

work page 2021

[8] [8]

E. J. Hu, Y . Shen, P. Wallis, Z. Allen-Zhu, Y . Li, S. Wang, L. Wang, and W. Chen. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685, 2021

arXiv 2021