Skip to main content

Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

  • Article
  • Published:

Versatile cardiovascular signal generation with a unified diffusion transformer

Nature Machine Intelligence volume 8, pages 6–19 (2026)Cite this article

Subjects

A preprint version of the article is available at arXiv.

Abstract

Cardiovascular signals such as photoplethysmography, electrocardiography and blood pressure are inherently correlated and complementary, together reflecting the health of the cardiovascular system. However, their joint utilization in real-time monitoring is severely limited by diverse acquisition challenges from noisy wearable recordings to burdened invasive procedures. Here we propose UniCardio, a multimodal diffusion transformer that reconstructs low-quality signals and synthesizes unrecorded signals in a unified generative framework. Its key innovations include a specialized model architecture to manage the signal modalities involved in generation tasks and a continual learning paradigm to incorporate varying modality combinations. By exploiting the complementary nature of cardiovascular signals, UniCardio clearly outperforms recent task-specific baselines in signal denoising, imputation and translation. The generated signals match the performance of ground-truth signals in detecting abnormal health conditions and estimating vital signs, even in unseen domains, as well as ensuring interpretability for human experts. These advantages establish UniCardio as a practical and robust framework for advancing artificial-intelligence-assisted healthcare.

This is a preview of subscription content, access via your institution

Access options

Buy this article

  • Purchase on SpringerLink
  • Instant access to the full article PDF.

USD 39.95

Prices may be subject to local taxes which are calculated during checkout

Fig. 1: Real-time monitoring and diagnosis of cardiovascular signals.
Fig. 2: Model architecture and training paradigm.
Fig. 3: Overall performance of versatile generation tasks.
Fig. 4: Versatile generation-assisted cardiovascular applications.
Fig. 5: Visualization of typical ECG abnormalities.

Similar content being viewed by others

Data availability

All benchmark datasets used in this paper are publicly available, including the cuffless BP dataset19 (https://archive.ics.uci.edu/dataset/340/cuff+less+blood+pressure+estimation), the PTB-XL dataset39 (https://physionet.org/content/ptb-xl/1.0.3), the MIMIC dataset42 (https://physionet.org/content/mimicdb/1.0.0), the MIMIC PERform AF dataset40 (https://ppg-beats.readthedocs.io/en/latest/datasets/mimic_perform_af) and the WESAD dataset41 (https://archive.ics.uci.edu/dataset/465/wesad+wearable+stress+and+affect+detection).

Code availability

The implementation code is available via Zenodo at https://doi.org/10.5281/zenodo.17240784 (ref. 79).

References

  1. WHO. Cardiovascular Diseases (CVDs) (WHO, 2021); https://www.who.int/health-topics/cardiovascular-diseases

  2. Alian, A. A. & Shelley, K. H. Photoplethysmography. Best Pract. Res. Clin. Anaesthesiol. 28, 395–406 (2014).

    Article  Google Scholar 

  3. Mirvis, D. M. & Goldberger, A. L. Electrocardiography. Heart Dis. 1, 82–128 (2001).

    Google Scholar 

  4. McGhee, B. H. & Bridges, E. J. Monitoring arterial blood pressure: what you may not know. Crit. Care Nurse 22, 60–79 (2002).

    Article  Google Scholar 

  5. Elgendi, M. et al. Recommendations for evaluating photoplethysmography-based algorithms for blood pressure assessment. Commun. Med. 4, 140 (2024).

    Article  Google Scholar 

  6. Tamura, T., Maeda, Y., Sekine, M. & Yoshida, M. Wearable photoplethysmographic sensors—past and present. Electronics 3, 282–302 (2014).

    Article  Google Scholar 

  7. Kligfield, P. et al. Recommendations for the standardization and interpretation of the electrocardiogram. Circulation 115, 1306–1324 (2007).

    Article  Google Scholar 

  8. Trobec, R., Tomašić, I., Rashkovska, A., Depolli, M. & Avbelj, V. Body Sensors and Electrocardiography (Springer, 2018).

  9. Pascual, J. L., Horak, J., Gracias, V. H. & Neligan, P. J. in Monitoring in Neurocritical Care (eds Le Roux et al.) 176–188 (Elsevier, 2013).

  10. Saugel, B., Dueck, R. & Wagner, J. Y. Measurement of blood pressure. Best Pract. Res. Clin. Anaesthesiol. 28, 309–322 (2014).

    Article  Google Scholar 

  11. Chiang, H.-T. et al. Noise reduction in ECG signals using fully convolutional denoising autoencoders. IEEE Access 7, 60806–60813 (2019).

    Article  Google Scholar 

  12. Ahmed, R., Mehmood, A., Rahman, M. M. U. & Dobre, O. A. A deep learning and fast wavelet transform-based hybrid approach for denoising of PPG signals. IEEE Sens. Lett. 7, 6003504 (2023).

    Article  Google Scholar 

  13. Xu, M. et al. PulseImpute: a novel benchmark task for pulsative physiological signal imputation. In 36th Conference on Neural Information Processing Systems (NeurIPS 2022) Track on Datasets and Benchmarks https://papers.neurips.cc/paper_files/paper/2022/file/ac01e21bb14609416760f790dd8966ae-Paper-Datasets_and_Benchmarks.pdf (2022).

  14. Bansal, P., Deshpande, P. & Sarawagi, S. Missing value imputation on multidimensional time series. Proc. VLDB Endow. 14, 2533–2545 (2021).

    Article  Google Scholar 

  15. Shome, D., Sarkar, P. & Etemad, A. Region-disentangled diffusion model for high-fidelity PPG-to-ECG translation. In Proc. AAAI Conference on Artificial Intelligence 15009–15019 (AAAI Press, 2024).

  16. Sarkar, P. & Etemad, A. CardioGAN: attentive generative adversarial network with dual discriminators for synthesis of ECG from PPG. In Proc. AAAI Conference on Artificial Intelligence 488–496 (AAAI Press, 2021).

  17. Goodfellow, I. et al. Generative adversarial networks. Commun. ACM 63, 139–144 (2020).

    Article  Google Scholar 

  18. De Bortoli, V., Thornton, J., Heng, J. & Doucet, A. Diffusion Schrödinger bridge with applications to score-based generative modeling. Adv. Neural Inf. Process. Syst. 34, 17695–17709 (2021).

    Google Scholar 

  19. Kachuee, M., Kiani, M. M., Mohammadzade, H. & Shabany, M. Cuffless blood pressure estimation algorithms for continuous health-care monitoring. IEEE Trans. Biomed. Eng. 64, 859–869 (2016).

    Article  Google Scholar 

  20. Ho, J., Jain, A. & Abbeel, P. Denoising diffusion probabilistic models. In 34th Conference on Neural Information Processing Systems (NeurIPS 2020) https://proceedings.neurips.cc/paper/2020/file/4c5bcfec8584af0d967f1ab10179ca4b-Paper.pdf (2020).

  21. Song, Y. et al. Score-based generative modeling through stochastic differential equations. In Proc. International Conference on Learning Representations https://openreview.net/pdf/ef0eadbe07115b0853e964f17aa09d811cd490f1.pdf?ref=news-tutorials-ai-research (ICLR, 2021).

  22. Bao, F. et al. One transformer fits all distributions in multi-modal diffusion at scale. In Proc. International Conference on Machine Learning 1692–1717 (PMLR, 2023).

  23. Peebles, W. & Xie, S. Scalable diffusion models with transformers. In Proc. IEEE/CVF International Conference on Computer Vision 4195–4205 (IEEE, 2023).

  24. Bao, F. et al. All are worth words: a ViT backbone for diffusion models. In Proc. IEEE/CVF Conference on Computer Vision and Pattern Recognition 22669–22679 (IEEE, 2023).

  25. Wang, L., Zhang, X., Su, H. & Zhu, J. A comprehensive survey of continual learning: theory, method and application. IEEE Trans. Pattern Anal. Mach. Intell. 46, 5362–5383 (2024).

    Article  Google Scholar 

  26. Parisi, G. I., Kemker, R., Part, J. L., Kanan, C. & Wermter, S. Continual lifelong learning with neural networks: a review. Neural Netw. 113, 54–71 (2019).

    Article  Google Scholar 

  27. Kirkpatrick, J. et al. Overcoming catastrophic forgetting in neural networks. Proc. Natl Acad. Sci. USA 114, 3521–3526 (2017).

    Article  MathSciNet  Google Scholar 

  28. Wang, L. et al. AFEC: active forgetting of negative transfer in continual learning. Adv. Neural Inf. Process. Syst. 34, 22379–22391 (2021).

    Google Scholar 

  29. Rebuffi, S.-A., Kolesnikov, A., Sperl, G. & Lampert, C. H. iCaRL: incremental classifier and representation learning. In Proc. IEEE Conference on Computer Vision and Pattern Recognition 2001–2010 (IEEE, 2017).

  30. Wang, L. et al. Memory replay with data compression for continual learning. In Proc. International Conference on Learning Representations https://openreview.net/pdf?id=a7H7OucbWaU (ICLR, 2021).

  31. Serra, J., Suris, D., Miron, M. & Karatzoglou, A. Overcoming catastrophic forgetting with hard attention to the task. In Proc. International Conference on Machine Learning 4548–4557 (PMLR, 2018).

  32. Wang, L. et al. Incorporating neuro-inspired adaptability for continual learning in artificial intelligence. Nature Mach. Intell. 5, 1356–1368 (2023).

    Article  Google Scholar 

  33. Bing, P. et al. A novel approach for denoising electrocardiogram signals to detect cardiovascular diseases using an efficient hybrid scheme. Front. Cardiovasc. Med. 11, 1277123 (2024).

    Article  Google Scholar 

  34. Paviglianiti, A., Randazzo, V., Pasero, E. & Vallan, A. Noninvasive arterial blood pressure estimation using ABPNet and VITAL-ECG. In Proc. IEEE International Instrumentation and Measurement Technology Conference 1–5 (IEEE, 2020).

  35. Ibtehaz, N. et al. PPG2ABP: translating photoplethysmogram (PPG) signals to arterial blood pressure (ABP) waveforms. Bioengineering 9, 692 (2022).

    Article  Google Scholar 

  36. Sendur, L. & Selesnick, I. W. Bivariate shrinkage with local variance estimation. IEEE Signal Process. Lett. 9, 438–441 (2003).

    Article  Google Scholar 

  37. Gerchberg, R. Super-resolution through error energy reduction. Opt. Acta 21, 709–720 (1974).

    Article  Google Scholar 

  38. Papoulis, A. A new algorithm in spectral analysis and band-limited extrapolation. IEEE Trans. Circuits Syst. 22, 735–742 (2003).

    Article  MathSciNet  Google Scholar 

  39. Wagner, P. et al. PTB-XL, a large publicly available electrocardiography dataset. Sci. Data 7, 154 (2020).

    Article  Google Scholar 

  40. Charlton, P. H. et al. Detecting beats in the photoplethysmogram: benchmarking open-source algorithms. Physiol. Meas. 43, 085007 (2022).

    Article  Google Scholar 

  41. Schmidt, P., Reiss, A., Duerichen, R., Marberger, C. & Van Laerhoven, K. Introducing WESAD, a multimodal dataset for wearable stress and affect detection. In Proc. ACM International Conference on Multimodal Interaction 400–408 (ACM, 2018).

  42. Moody, G. B. & Mark, R. G. A database to support development and evaluation of intelligent intensive care monitoring. In Proc. Computers in Cardiology 657–660 (IEEE, 1996).

  43. Finnegan, E. et al. Features from the photoplethysmogram and the electrocardiogram for estimating changes in blood pressure. Sci. Rep. 13, 986 (2023).

    Article  Google Scholar 

  44. Landry, C. & Mukkamala, R. Current evidence suggests that estimating blood pressure from convenient ECG waveforms alone is not viable. J. Electrocardiol. 81, 153–155 (2023).

    Article  Google Scholar 

  45. Kirchhof, P. et al. 2016 ESC guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Kardiol. Pol. 74, 1359–1469 (2016).

    Article  Google Scholar 

  46. January, C. T. et al. 2014 AHA/ACC/HRS guideline for the management of patients with atrial fibrillation: executive summary: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines and the Heart Rhythm Society. Circulation 130, 2071–2104 (2014).

    Article  Google Scholar 

  47. Zipes, D. P., Jalife, J. & Stevenson, W. G. Cardiac Electrophysiology: From Cell to Bedside E-Book (Elsevier Health Sciences, 2017).

  48. Thygesen, K. et al. Fourth universal definition of myocardial infarction.Circulation 138, 618–651 (2018).

    Article  Google Scholar 

  49. Antman, E. M. et al. The TIMI risk score for unstable angina/non-ST elevation MI: a method for prognostication and therapeutic decision making. JAMA 284, 835–842 (2000).

    Article  Google Scholar 

  50. Amsterdam, E. A. et al. 2014 AHA/ACC guideline for the management of patients with non–ST-elevation acute coronary syndromes: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J. Am. Coll. Cardiol. 64, 139–228 (2014).

    Article  Google Scholar 

  51. Karthikeyan, P., Murugappan, M. & Yaacob, S. A review on stress inducement stimuli for assessing human stress using physiological signals. In Proc. IEEE International Colloquium on Signal Processing and Its Applications 420–425 (IEEE, 2011).

  52. de Santos Sierra, A., Ávila, C. S., Casanova, J. G. & Del Pozo, G. B. A stress-detection system based on physiological signals and fuzzy logic. IEEE Trans. Ind. Electron. 58, 4857–4865 (2011).

    Article  Google Scholar 

  53. Momeni, N., Dell’Agnola, F., Arza, A. & Atienza, D. Real-time cognitive workload monitoring based on machine learning using physiological signals in rescue missions. In Proc. Annual International Conference of the IEEE Engineering in Medicine and Biology Society 3779–3785 (IEEE, 2019).

  54. Ranchet, M., Morgan, J. C., Akinwuntan, A. E. & Devos, H. Cognitive workload across the spectrum of cognitive impairments: a systematic review of physiological measures. Neurosci. Biobehav. Rev. 80, 516–537 (2017).

    Article  Google Scholar 

  55. Shu, L. et al. A review of emotion recognition using physiological signals. Sensors 18, 2074 (2018).

    Article  Google Scholar 

  56. Ayata, D., Yaslan, Y. & Kamasak, M. E. Emotion recognition from multimodal physiological signals for emotion aware healthcare systems. J. Med. Biol. Eng. 40, 149–157 (2020).

    Article  Google Scholar 

  57. Xuan, Q. et al. Assessing cognitive load in adolescent and adult students using photoplethysmogram morphometrics. Cogn. Neurodyn. 14, 709–721 (2020).

    Article  Google Scholar 

  58. Pavlov, Y. G. et al. Task-evoked pulse wave amplitude tracks cognitive load. Sci. Rep. 13, 22592 (2023).

    Article  Google Scholar 

  59. Al Abdi, R. M., Alhitary, A. E., Abdul Hay, E. W. & Al-Bashir, A. K. Objective detection of chronic stress using physiological parameters. Med. Biol. Eng. Comput. 56, 2273–2286 (2018).

    Article  Google Scholar 

  60. Mandrick, K., Peysakhovich, V., Rémy, F., Lepron, E. & Causse, M. Neural and psychophysiological correlates of human performance under stress and high mental workload. Biol. Psychol. 121, 62–73 (2016).

    Article  Google Scholar 

  61. Richter, M., Friedrich, A. & Gendolla, G. H. Task difficulty effects on cardiac activity. Psychophysiology 45, 869–875 (2008).

    Article  Google Scholar 

  62. El Sayed, K., Macefield, V. G., Hissen, S. L., Joyner, M. J. & Taylor, C. E. Rate of rise in diastolic blood pressure influences vascular sympathetic response to mental stress. J. Physiol. 594, 7465–7482 (2016).

    Article  Google Scholar 

  63. Cao, Y. et al. A survey of AI-generated content (AIGC). ACM Comput. Surv. https://doi.org/10.1145/3704262 (2025).

  64. Mai, W., Zhang, J., Fang, P. & Zhang, Z. Brain-conditional multimodal synthesis: a survey and taxonomy. IEEE Trans. Artif. Intell. 6, 1080–1099 (2025).

    Article  Google Scholar 

  65. Elgendi, M. Optimal signal quality index for photoplethysmogram signals. Bioengineering 3, 21 (2016).

    Article  Google Scholar 

  66. Zhu, F., Ye, F., Fu, Y., Liu, Q. & Shen, B. Electrocardiogram generation with a bidirectional LSTM-CNN generative adversarial network. Sci. Rep. 9, 6734 (2019).

    Article  Google Scholar 

  67. Mazumder, O. et al. Synthetic PPG signal generation to improve coronary artery disease classification: study with physical model of cardiovascular system. IEEE J. Biomed. Health Inform. 26, 2136–2146 (2022).

    Article  Google Scholar 

  68. Ezzat, A., Omer, O. A., Mohamed, U. S. & Mubarak, A. S. ECG signal reconstruction from PPG using a hybrid attention-based deep learning network. EURASIP J. Adv. Signal Process. 2024, 95 (2024).

    Article  Google Scholar 

  69. van den Oord, A. et al. Conditional image generation with PixelCNN decoders. In Proc. Advances in Neural Information Processing Systems https://proceedings.neurips.cc/paper_files/paper/2016/file/b1301141feffabac455e1f90a7de2054-Paper.pdf (2016).

  70. van den Oord, A. et al. WaveNet: a generative model for raw audio. Preprint at https://arxiv.org/abs/1609.03499 (2016).

  71. Kong, Z., Ping, W., Huang, J., Zhao, K. & Catanzaro, B. DiffWave: a versatile diffusion model for audio synthesis. In Proc. International Conference on Learning Representations https://openreview.net/pdf?id=a-xFK8Ymz5J (ICLR, 2021).

  72. Tashiro, Y., Song, J., Song, Y. & Ermon, S. CSDI: conditional score-based diffusion models for probabilistic time series imputation. In 35th Conference on Neural Information Processing Systems (NeurIPS 2021) https://papers.neurips.cc/paper_files/paper/2021/file/cfe8504bda37b575c70ee1a8276f3486-Paper.pdf (2021).

  73. Makowski, D. et al. NeuroKit2: a Python toolbox for neurophysiological signal processing. Behav. Res. Methods 53, 1689–1696 (2021).

  74. Richman, J. S. & Moorman, J. R. Physiological time-series analysis using approximate entropy and sample entropy. Am. J. Physiol. Heart Circ. Physiol. 278, 2039–2049 (2000).

    Article  Google Scholar 

  75. Liu, Z. et al. Multiclass arrhythmia detection and classification from photoplethysmography signals using a deep convolutional neural network. J. Am. Heart Assoc. 11, 023555 (2022).

    Article  Google Scholar 

  76. Nabian, M. et al. An open-source feature extraction tool for the analysis of peripheral physiological data. IEEE J. Transl. Eng. Health Med. 6, 2800711 (2018).

    Article  Google Scholar 

  77. Jeong, D. U. & Lim, K. M. Combined deep CNN–LSTM network-based multitasking learning architecture for noninvasive continuous blood pressure estimation using difference in ECG–PPG features. Sci. Rep. 11, 13539 (2021).

    Article  Google Scholar 

  78. NVIDIA Corporation. Jetson Benchmark Performance (2023); https://developer.nvidia.com/embedded/jetson-benchmarks

  79. Chen, Z., Miao, Y. & Wang, L. thu-ml/unicardio: UniCardio paper. Zenodo https://doi.org/10.5281/zenodo.8293564 (2025).

Download references

Acknowledgements

This work was supported by the NSFC Projects (62350080 to J.Z., 92270001 to J.Z., U24A20342 to Z.C. and 62406160 to L.W.), Tsinghua Institute for Guo Qiang and the High Performance Computing Center, Tsinghua University. J.Z. is also supported by the Xplorer Prize.

Author information

Author notes

  1. These authors contributed equally: Zehua Chen, Yuyang Miao, Liyuan Wang.

Authors and Affiliations

  1. Department of Computer Science and Technology, Institute for AI, BNRist Center, THBI Lab, Tsinghua-Bosch Joint Center for ML, Tsinghua University, Beijing, China

    Zehua Chen, Yuyang Miao, Liyuan Wang & Jun Zhu

  2. Department of Electrical and Electronic Engineering, Imperial College London, London, UK

    Yuyang Miao & Danilo P. Mandic

  3. Department of Psychological and Cognitive Sciences, Tsinghua University, Beijing, China

    Liyuan Wang

  4. Beijing Anzhen Hospital of Capital Medical University, Beijing Institute of Heart Lung and Blood Vessel Diseases, Chinese Institutes for Medical Research, Beijing, China

    Luyun Fan

Authors
  1. Zehua Chen
    View author publications

    Search author on:PubMed Google Scholar

  2. Yuyang Miao
    View author publications

    Search author on:PubMed Google Scholar

  3. Liyuan Wang
    View author publications

    Search author on:PubMed Google Scholar

  4. Luyun Fan
    View author publications

    Search author on:PubMed Google Scholar

  5. Danilo P. Mandic
    View author publications

    Search author on:PubMed Google Scholar

  6. Jun Zhu
    View author publications

    Search author on:PubMed Google Scholar

Contributions

Z.C., Y.M., L.W. and J.Z. conceived the project. Z.C., Y.M. and L.W. designed the computational model. Z.C. and Y.M. performed the main experiments, assisted by L.W. Z.C., Y.M. and L.W. analysed the data. L.W. wrote the paper, assisted by Z.C., Y.M. and L.F. Z.C., Y.M., L.W., L.F., D.P.M. and J.Z. revised the paper. L.W. and J.Z. supervised the project.

Corresponding authors

Correspondence to Liyuan Wang or Jun Zhu.

Ethics declarations

Competing interests

The authors declare no competing interests.

Peer review

Peer review information

Nature Machine Intelligence thanks Gianmarco Mengaldo and the other, anonymous, reviewer(s) for their contribution to the peer review of this work.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Supplementary Figs. 1–6 and Tables 1–5.

Reporting Summary

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Chen, Z., Miao, Y., Wang, L. et al. Versatile cardiovascular signal generation with a unified diffusion transformer. Nat Mach Intell 8, 6–19 (2026). https://doi.org/10.1038/s42256-025-01147-y

Download citation

  • Received:

  • Accepted:

  • Published:

  • Version of record:

  • Issue date:

  • DOI: https://doi.org/10.1038/s42256-025-01147-y

This article is cited by

Search

Advanced search

Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing