arxiv: 2605.12541 · v1 · submitted 2026-05-09 · 📡 eess.SP · cs.AI· cs.LG

Recognition: 2 theorem links

· Lean Theorem

PG-LRF: Physiology-Guided Latent Rectified Flow for Electro-Hemodynamic PPG-to-ECG Generation

Xiaoda Wang , Minxiao Wang , Kaiqiao Han , Defu Cao , Ching Chang , Yidan Shi , Runze Yan , Xiao Luo

show 5 more authors

Yan Liu Xiao Hu Yizhou Sun Wei Wang Carl Yang

Authors on Pith no claims yet

Pith reviewed 2026-05-14 21:47 UTC · model grok-4.3

classification 📡 eess.SP cs.AIcs.LG

keywords PPG-to-ECG generationrectified flowphysiology-guidedelectro-hemodynamic simulatorcardiac phase dynamicswearable monitoringsignal synthesiscardiovascular classification

0 comments

The pith

PG-LRF uses a shared cardiac-phase simulator to guide latent rectified flow and produce ECG signals from PPG that respect both signal details and hemodynamic rules.

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

The paper tries to recover full ECG morphology and timing from PPG signals captured by everyday wearables. It does so by first building an electro-hemodynamic simulator that links the two signals through common cardiac-phase variables, then training a latent rectified flow model whose transport steps are constrained to stay consistent with the simulator’s forward predictions. A sympathetic reader would care because this turns ubiquitous PPG readings into clinically richer ECG traces without extra hardware. The approach replaces purely statistical mapping with explicit physiological constraints on morphology and forward hemodynamics. Experiments on the MC-MED dataset show gains in both reconstruction accuracy and downstream disease classification.

Core claim

PG-LRF introduces an electro-hemodynamic simulator that co-models ECG and PPG through shared cardiac phase dynamics. Guided by this simulator, a Physiology-Aware AutoEncoder learns a structured electro-hemodynamic latent space. The simulator guidance is integrated into a PPG-conditioned latent rectified flow that enforces ECG-side morphology consistency and ECG-to-PPG forward hemodynamic consistency during generative transport. On the MC-MED dataset the resulting ECGs improve both signal fidelity and cardiovascular disease classification performance.

What carries the argument

Electro-hemodynamic simulator that co-models ECG and PPG via shared cardiac phase dynamics, used to structure the latent space and to enforce morphology plus forward-hemodynamic consistency inside the rectified-flow transport steps.

If this is right

Generated ECGs support more accurate cardiovascular disease classification than those from prior PPG-to-ECG methods.
The outputs remain consistent with the forward hemodynamic pathway from electrical activity to peripheral pulse.
The latent space becomes organized around physiologically meaningful factors instead of pure statistical correlations.
The same guidance mechanism can be applied to other conditional generation tasks that involve paired physiological signals.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Consumer wearables could run this model locally to deliver ECG-level diagnostics from continuous PPG streams.
The simulator-plus-rectified-flow pattern may transfer to other cross-modal medical signal tasks such as EEG-to-MEG or fNIRS-to-EEG synthesis.
Extending the simulator to include additional factors like respiration or posture could further improve robustness on noisy ambulatory recordings.

Load-bearing premise

The simulator correctly captures the true shared cardiac-phase relationship between ECG and PPG, so that enforcing consistency during generation yields real physiological signals rather than simulator artifacts.

What would settle it

On an independent paired ECG-PPG test set recorded with different sensors or patient groups, the generated ECGs would show mismatched timing or morphology relative to the simultaneously measured true ECG.

Figures

Figures reproduced from arXiv: 2605.12541 by Carl Yang, Ching Chang, Defu Cao, Kaiqiao Han, Minxiao Wang, Runze Yan, Wei Wang, Xiaoda Wang, Xiao Hu, Xiao Luo, Yan Liu, Yidan Shi, Yizhou Sun.

**Figure 1.** Figure 1: Overview Framework of PG-LRF. (a) The Electro-Hemodynamic Physiological Simulator: co-models ECG and PPG through shared cardiac phase dynamics. (b) Physiology-Aware AutoEncoder: learns an aligned electro-hemodynamic latent space from native-rate PPG and ECG. (c) Simulator-Guided Latent Rectified Flow: generates ECG latents conditioned on PPG and regularizes the transport with ECG morphology and ECG-to-PPG… view at source ↗

**Figure 2.** Figure 2: Qualitative case studies of PPG-to-ECG generation by PG-LRF. Across the three cases, [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗

read the original abstract

Electrocardiography (ECG) is the clinical standard for cardiac assessment but requires dedicated hardware that does not scale to daily-life monitoring. Photoplethysmography (PPG) is ubiquitous in wearables but lacks ECG-specific diagnostic morphology and is corrupted by motion and sensor noise. PPG-to-ECG generation aims to bridge this gap by recovering electrical morphology and timing from peripheral pulse signals. However, existing methods largely rely on statistical alignment and data-driven generation. They fail to explicitly structure the latent space around physiology-aware electro-hemodynamic factors and lack constraints from forward physiological dynamics. To address these challenges, we propose PG-LRF, a physiology-guided latent rectified flow framework. PG-LRF introduces an electro-hemodynamic simulator that co-models ECG and PPG through shared cardiac phase dynamics. Guided by this simulator, a Physiology-Aware AutoEncoder learns a structured electro-hemodynamic latent space. Then we integrate this simulator guidance into a PPG-conditioned latent rectified flow, enforcing ECG-side morphology consistency and ECG-to-PPG forward hemodynamic consistency during generative transport. Experiments on the large-scale MC-MED dataset demonstrate that PG-LRF significantly improves PPG-to-ECG generation and downstream cardiovascular disease classification, proving its ability to generate ECGs that are both signal-faithful and physiologically plausible under the ECG-to-PPG hemodynamic pathway

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

PG-LRF adds simulator-guided consistency enforcement to latent rectified flows for PPG-to-ECG translation, but the claims rest on an unvalidated electro-hemodynamic model.

read the letter

The paper's main new piece is the PG-LRF framework, which uses an electro-hemodynamic simulator to structure the latent space in a physiology-aware autoencoder and then folds that same simulator into a PPG-conditioned rectified flow. During transport it adds explicit checks for ECG morphology consistency and ECG-to-PPG forward hemodynamic consistency. That specific combination of simulator guidance at both the autoencoding and flow stages is not something I have seen laid out in the prior PPG-to-ECG literature referenced here. The large MC-MED dataset and the downstream cardiovascular classification task are reasonable choices for showing practical impact. The framing also correctly identifies that most existing translation methods stay purely statistical and do not bake in forward physiological dynamics. Those are the parts that feel like genuine forward steps. The soft spots are straightforward. The abstract supplies no numbers, baselines, ablations, or error bars, so the size of any improvement is impossible to judge from what is shown. More critically, the entire argument depends on the simulator accurately co-modeling real ECG-PPG pairs through shared cardiac phase dynamics. No independent validation of the simulator against external measurements is mentioned, which leaves open the possibility that the consistency constraints are simply locking in simulator biases rather than real physiology. That circularity risk is load-bearing and needs direct evidence before the physiological-plausibility claim can be taken at face value. This work is aimed at the biomedical signal-processing and wearable-AI community. Readers already working on generative models for physiological time series would get the most out of the framework details, assuming the full experiments hold up. I would send it to peer review. The core idea is concrete enough to deserve referee scrutiny on the simulator validation and quantitative results, even if revisions are required.

Referee Report

2 major / 1 minor

Summary. The paper proposes PG-LRF, a physiology-guided latent rectified flow framework for PPG-to-ECG generation. It introduces an electro-hemodynamic simulator that co-models ECG and PPG through shared cardiac phase dynamics, a Physiology-Aware AutoEncoder to learn a structured latent space, and integrates simulator guidance into a PPG-conditioned latent rectified flow enforcing morphology and forward hemodynamic consistency. Experiments on the MC-MED dataset are claimed to show significant improvements in generation quality and downstream cardiovascular disease classification, demonstrating signal-faithful and physiologically plausible ECG outputs.

Significance. If the results hold and the simulator is validated independently, this work could have high significance for scalable cardiac monitoring using wearables, bridging the gap between ubiquitous PPG and diagnostic ECG morphology. The integration of physiological constraints into generative models is a promising direction, but the current presentation leaves the magnitude of improvement and robustness unclear.

major comments (2)

[Abstract] The abstract asserts 'significant improvements' in PPG-to-ECG generation and downstream classification on the MC-MED dataset without providing any quantitative metrics, baseline comparisons, error bars, or ablation results. This makes the central claim difficult to evaluate and requires the full experimental section to include these details for assessment.
[Abstract] The electro-hemodynamic simulator is central to structuring the latent space and enforcing consistency constraints, yet no independent validation metrics against real physiological measurements are referenced. If the simulator contains biases in phase dynamics or forward mapping, the consistency enforcement may embed artifacts rather than true physiology, as the constraints are defined w.r.t. the same simulator.

minor comments (1)

The term 'Physiology-Aware AutoEncoder' is introduced without a clear definition or architectural details in the abstract; this should be expanded in the methods section.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment point by point below, providing clarifications and committing to revisions where appropriate to strengthen the presentation of results and simulator validation.

read point-by-point responses

Referee: [Abstract] The abstract asserts 'significant improvements' in PPG-to-ECG generation and downstream classification on the MC-MED dataset without providing any quantitative metrics, baseline comparisons, error bars, or ablation results. This makes the central claim difficult to evaluate and requires the full experimental section to include these details for assessment.

Authors: We agree that the abstract would be strengthened by including key quantitative results to support the claims of significant improvements. The full experimental section already reports detailed metrics including MSE reductions of over 30% relative to baselines, DTW alignment scores, downstream classification accuracy gains with error bars and statistical tests, plus ablation studies. In the revised manuscript we will update the abstract to concisely incorporate the most salient numerical results (e.g., specific error reductions and p-values) while preserving length constraints. revision: yes
Referee: [Abstract] The electro-hemodynamic simulator is central to structuring the latent space and enforcing consistency constraints, yet no independent validation metrics against real physiological measurements are referenced. If the simulator contains biases in phase dynamics or forward mapping, the consistency enforcement may embed artifacts rather than true physiology, as the constraints are defined w.r.t. the same simulator.

Authors: We acknowledge the importance of independent simulator validation to rule out potential circularity or bias. The simulator is grounded in established physiological models from the literature and its parameters are calibrated on real MC-MED data. In the revision we will add an explicit validation subsection (or appendix) reporting quantitative comparisons of simulated versus real ECG-PPG pairs, including phase correlation coefficients, waveform RMSE, and hemodynamic consistency metrics. We will also expand the discussion to address how data-driven components in the rectified flow mitigate any residual simulator biases. revision: yes

Circularity Check

1 steps flagged

Simulator-defined physiology makes consistency constraints self-referential

specific steps

self definitional [Abstract]
"PG-LRF introduces an electro-hemodynamic simulator that co-models ECG and PPG through shared cardiac phase dynamics. Guided by this simulator, a Physiology-Aware AutoEncoder learns a structured electro-hemodynamic latent space. Then we integrate this simulator guidance into a PPG-conditioned latent rectified flow, enforcing ECG-side morphology consistency and ECG-to-PPG forward hemodynamic consistency during generative transport."

The morphology consistency and forward hemodynamic consistency are defined and enforced using the identical simulator that structures the latent space; therefore the assertion of physiological plausibility is equivalent to consistency with the paper's own simulator model by construction.

full rationale

The paper introduces its own electro-hemodynamic simulator to co-model ECG and PPG via shared cardiac phase dynamics, then uses that same simulator both to structure the latent space in the Physiology-Aware AutoEncoder and to enforce morphology and forward hemodynamic consistency inside the rectified flow. The central claim that generated ECGs are 'signal-faithful and physiologically plausible under the ECG-to-PPG hemodynamic pathway' therefore reduces to internal consistency with the simulator's own assumptions rather than external physiological measurements. This is a self-definitional reduction: the 'physiology' that validates the outputs is defined by the model that generates them. No independent simulator validation metrics are referenced, so downstream gains on MC-MED risk being partly tautological to the simulator's construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

The central claim rests on the domain assumption that ECG and PPG share accurately modelable cardiac phase dynamics and that simulator-derived consistency constraints improve generation without introducing new artifacts.

axioms (1)

domain assumption ECG and PPG can be co-modeled through shared cardiac phase dynamics by an electro-hemodynamic simulator
Invoked to guide the Physiology-Aware AutoEncoder and to enforce consistency in the rectified flow.

invented entities (1)

Physiology-Aware AutoEncoder no independent evidence
purpose: Learns a structured electro-hemodynamic latent space
New component introduced to organize representations around physiology-aware factors.

pith-pipeline@v0.9.0 · 5581 in / 1363 out tokens · 86447 ms · 2026-05-14T21:47:03.793187+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

electro-hemodynamic simulator that co-models ECG and PPG through shared cardiac phase dynamics... dx/dt = α(x,y)x−ωy, dy/dt=α(x,y)y+ωx with α=1−sqrt(x²+y²)
IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

simulator-guided residual losses Le_sim and Lp_sim... Euler residual rm_ℓ = (˜hm,ℓ+1−˜hm,ℓ)/Δtm − fm(...)

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

34 extracted references · 34 canonical work pages · 1 internal anchor

[1]

Photoplethysmography and its application in clinical physiological measurement

John Allen. Photoplethysmography and its application in clinical physiological measurement. Physiological measurement, 28(3):R1–R39, 2007

work page 2007
[2]

A survey on ecg analysis.Biomedical Signal Processing and Control, 43:216–235, 2018

Selcan Kaplan Berkaya, Alper Kursat Uysal, Efnan Sora Gunal, Semih Ergin, Serkan Gunal, and M Bilginer Gulmezoglu. A survey on ecg analysis.Biomedical Signal Processing and Control, 43:216–235, 2018

work page 2018
[3]

A review on wearable photoplethysmography sensors and their potential future applications in health care.International journal of biosensors & bioelectronics, 4(4):195, 2018

Denisse Castaneda, Aibhlin Esparza, Mohammad Ghamari, Cinna Soltanpur, and Homer Nazeran. A review on wearable photoplethysmography sensors and their potential future applications in health care.International journal of biosensors & bioelectronics, 4(4):195, 2018

work page 2018
[4]

Wearable photoplethysmography for cardiovascular monitoring.Proceed- ings of the IEEE, 110(3):355–381, 2022

Peter H Charlton, Panicos A Kyriacou, Jonathan Mant, Vaidotas Marozas, Phil Chowienczyk, and Jordi Alastruey. Wearable photoplethysmography for cardiovascular monitoring.Proceed- ings of the IEEE, 110(3):355–381, 2022

work page 2022
[5]

Ricky T. Q. Chen, Yulia Rubanova, Jesse Bettencourt, and David Duvenaud. Neural ordinary differential equations. InAdvances in Neural Information Processing Systems (NeurIPS), 2018

work page 2018
[6]

Diffusion models in vision: A survey.IEEE transactions on pattern analysis and machine intelligence, 45(9):10850–10869, 2023

Florinel-Alin Croitoru, Vlad Hondru, Radu Tudor Ionescu, and Mubarak Shah. Diffusion models in vision: A survey.IEEE transactions on pattern analysis and machine intelligence, 45(9):10850–10869, 2023

work page 2023
[7]

Ai modeling photoplethysmography to electrocardiography useful for predicting cardiovascular disease.npj Digital Medicine, 2025

Zhengyao Ding, Yujian Hu, Ziyu Li, Yiheng Mao, Haitao Li, Dongchen Zhou, Xuesen Chu, Long Yu, Ziyi Liu, Fei Wu, et al. Ai modeling photoplethysmography to electrocardiography useful for predicting cardiovascular disease.npj Digital Medicine, 2025

work page 2025
[8]

An ecg simulator with a novel ecg profile for physiological signals.Journal of medical engineering & technology, 42(7):501–509, 2018

Jan-Christoph Edelmann, Dominik Mair, Daniel Ziesel, Martin Burtscher, and Thomas Uss- mueller. An ecg simulator with a novel ecg profile for physiological signals.Journal of medical engineering & technology, 42(7):501–509, 2018

work page 2018
[9]

Scaling rectified flow trans- formers for high-resolution image synthesis

Patrick Esser, Sumith Kulal, Andreas Blattmann, Rahim Entezari, Jonas Müller, Harry Saini, Yam Levi, Dominik Lorenz, Axel Sauer, Frederic Boesel, et al. Scaling rectified flow trans- formers for high-resolution image synthesis. InForty-first international conference on machine learning, 2024

work page 2024
[10]

Ppgflowecg: Latent rectified flow with cross-modal encoding for ppg-guided ecg generation and cardiovascular disease detection.arXiv preprint arXiv:2509.19774, 2025

Xiaocheng Fang, Jiarui Jin, Haoyu Wang, Che Liu, Jieyi Cai, Yujie Xiao, Guangkun Nie, Bo Liu, Shun Huang, Hongyan Li, et al. Ppgflowecg: Latent rectified flow with cross-modal encoding for ppg-guided ecg generation and cardiovascular disease detection.arXiv preprint arXiv:2509.19774, 2025

work page arXiv 2025
[11]

Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring.Biosensors, 11(4):126, 2021

Jesse Fine, Kimberly L Branan, Andres J Rodriguez, Tananant Boonya-Ananta, Ajmal, Jes- sica C Ramella-Roman, Michael J McShane, and Gerard L Cote. Sources of inaccuracy in photoplethysmography for continuous cardiovascular monitoring.Biosensors, 11(4):126, 2021

work page 2021
[12]

Denoising diffusion probabilistic models.Advances in neural information processing systems, 33:6840–6851, 2020

Jonathan Ho, Ajay Jain, and Pieter Abbeel. Denoising diffusion probabilistic models.Advances in neural information processing systems, 33:6840–6851, 2020

work page 2020
[13]

Multimodal clinical monitoring in the emergency department (mc-med).Published online March, 3, 2025

Aman Kansal, Emma Chen, Tom Jin, Pranav Rajpurkar, and David Kim. Multimodal clinical monitoring in the emergency department (mc-med).Published online March, 3, 2025

work page 2025
[14]

William Hancock, Gerard van Herpen, Jan A Kors, Peter Macfarlane, David M Mirvis, and et al

Paul Kligfield, Leonard S Gettes, James J Bailey, Rory Childers, Barbara J Deal, E. William Hancock, Gerard van Herpen, Jan A Kors, Peter Macfarlane, David M Mirvis, and et al. Recommendations for the standardization and interpretation of the electrocardiogram: Part i: The electrocardiogram and its technology.Circulation, 115(10):1306–1324, 2007

work page 2007
[15]

Performer: A novel ppg-to-ecg reconstruction transformer for a digital biomarker of cardiovascular disease detection

Ella Lan. Performer: A novel ppg-to-ecg reconstruction transformer for a digital biomarker of cardiovascular disease detection. InProceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pages 1991–1999, 2023

work page 1991
[16]

Improving the training of rectified flows.Advances in neural information processing systems, 37:63082–63109, 2024

Sangyun Lee, Zinan Lin, and Giulia Fanti. Improving the training of rectified flows.Advances in neural information processing systems, 37:63082–63109, 2024. 10

work page 2024
[17]

Clep-gan: an innovative approach to subject-independent ecg reconstruction from ppg signals

Xiaoyan Li, Shixin Xu, Faisal Habib, Neda Aminnejad, Arvind Gupta, and Huaxiong Huang. Clep-gan: an innovative approach to subject-independent ecg reconstruction from ppg signals. BMC bioinformatics, 26(1):306, 2025

work page 2025
[18]

Eddm: A novel ecg denoising method using dual-path diffusion model.IEEE Transactions on Instrumentation and Measurement, 2025

Zhiyuan Li, Yuanyuan Tian, Yanrui Jin, Xiaoyang Wei, Mengxiao Wang, Jinlei Liu, and Chengliang Liu. Eddm: A novel ecg denoising method using dual-path diffusion model.IEEE Transactions on Instrumentation and Measurement, 2025

work page 2025
[19]

Flow matching for generative modeling

Yaron Lipman, Ricky TQ Chen, Heli Ben-Hamu, Maximilian Nickel, and Matthew Le. Flow matching for generative modeling. InThe Eleventh International Conference on Learning Representations, 2023

work page 2023
[20]

A dynamical model for generating synthetic electrocardiogram signals.IEEE transactions on biomedical engineering, 50(3):289–294, 2003

Patrick E McSharry, Gari D Clifford, Lionel Tarassenko, and Leonard A Smith. A dynamical model for generating synthetic electrocardiogram signals.IEEE transactions on biomedical engineering, 50(3):289–294, 2003

work page 2003
[21]

Playing Atari with Deep Reinforcement Learning

V olodymyr Mnih, Koray Kavukcuoglu, David Silver, Alex Graves, Ioannis Antonoglou, Daan Wierstra, and Martin Riedmiller. Playing atari with deep reinforcement learning.arXiv preprint arXiv:1312.5602, 2013

work page internal anchor Pith review Pith/arXiv arXiv 2013
[22]

Cardioflow: Learning to generate ecg from ppg with rectified flow

Yuta Nambu, Masahiro Kohjima, and Ryuji Yamamoto. Cardioflow: Learning to generate ecg from ppg with rectified flow. InICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 1–5. IEEE, 2025

work page 2025
[23]

Physdiff: physiology-based dynamicity disentangled diffusion model for remote phys- iological measurement

Wei Qian, Gaoji Su, Dan Guo, Jinxing Zhou, Xiaobai Li, Bin Hu, Shengeng Tang, and Meng Wang. Physdiff: physiology-based dynamicity disentangled diffusion model for remote phys- iological measurement. InProceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 6568–6576, 2025

work page 2025
[24]

Universal differential equations for scientific machine learning.arXiv preprint arXiv:2001.04385, 2020

Christopher Rackauckas, Yingbo Ma, Julius Martensen, Collin Warner, Kirill Zubov, Rohit Supekar, Dominic Skinner, Ali Ramadhan, and Alan Edelman. Universal differential equations for scientific machine learning.arXiv preprint arXiv:2001.04385, 2020

work page arXiv 2001
[25]

High- resolution image synthesis with latent diffusion models

Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Björn Ommer. High- resolution image synthesis with latent diffusion models. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 10684–10695, 2022

work page 2022
[26]

Deep conditional generative model for personalization of 12-lead electrocardiograms and cardiovascular risk prediction

Yuling Sang, Abhirup Banerjee, Marcel Beetz, and Vicente Grau. Deep conditional generative model for personalization of 12-lead electrocardiograms and cardiovascular risk prediction. Frontiers in Digital Health, 7:1558589, 2025

work page 2025
[27]

Cardiogan: Attentive generative adversarial network with dual discriminators for synthesis of ecg from ppg

Pritam Sarkar and Ali Etemad. Cardiogan: Attentive generative adversarial network with dual discriminators for synthesis of ecg from ppg. InProceedings of the AAAI Conference on Artificial Intelligence, volume 35, pages 488–496, 2021

work page 2021
[28]

Region-disentangled diffusion model for high-fidelity ppg-to-ecg translation

Debaditya Shome, Pritam Sarkar, and Ali Etemad. Region-disentangled diffusion model for high-fidelity ppg-to-ecg translation. InProceedings of the AAAI conference on artificial intelligence, volume 38, pages 15009–15019, 2024

work page 2024
[29]

Contrastive flow matching

George Stoica, Vivek Ramanujan, Xiang Fan, Ali Farhadi, Ranjay Krishna, and Judy Hoffman. Contrastive flow matching. InProceedings of the IEEE/CVF International Conference on Computer Vision, pages 1185–1194, 2025

work page 2025
[30]

Synthetic photoplethys- mogram generation using two gaussian functions.Scientific Reports, 10(1):13883, 2020

Qunfeng Tang, Zhencheng Chen, Rabab Ward, and Mohamed Elgendi. Synthetic photoplethys- mogram generation using two gaussian functions.Scientific Reports, 10(1):13883, 2020

work page 2020
[31]

Qunfeng Tang, Zhencheng Chen, Rabab Ward, Carlo Menon, and Mohamed Elgendi. Ppg2ecgps: An end-to-end subject-specific deep neural network model for electrocardiogram reconstruction from photoplethysmography signals without pulse arrival time adjustments.Bioengineering, 10(6):630, 2023

work page 2023
[32]

Position: Beyond prediction: Toward verifiable physiological waveform reasoning with foundation models and agentic llms

Xiaoda Wang, Ching Chang, Defu Cao, Kaiqiao Han, Fang Sun, Yue Huang, Minxiao Wang, Chang Xu, Xiao Luo, Runze Yan, et al. Position: Beyond prediction: Toward verifiable physiological waveform reasoning with foundation models and agentic llms. InThe Forty-Third International Conference on Machine Learning, 2026. 11

work page 2026
[33]

Flow matching for scalable simulation-based inference.Advances in Neural Information Processing Systems, 36:16837–16864, 2023

Jonas Wildberger, Maximilian Dax, Simon Buchholz, Stephen Green, Jakob H Macke, and Bernhard Schölkopf. Flow matching for scalable simulation-based inference.Advances in Neural Information Processing Systems, 36:16837–16864, 2023

work page 2023
[34]

Diffusion models: A comprehensive survey of methods and applications.ACM computing surveys, 56(4):1–39, 2023

Ling Yang, Zhilong Zhang, Yang Song, Shenda Hong, Runsheng Xu, Yue Zhao, Wentao Zhang, Bin Cui, and Ming-Hsuan Yang. Diffusion models: A comprehensive survey of methods and applications.ACM computing surveys, 56(4):1–39, 2023. 12 A Related Work A.1 Latent Generative Models Diffusion and score-based generative models have become powerful tools for learning...

work page 2023