Key result
A time-series-to-time-series generative adversarial network (T2TGAN) estimated continuous mean arterial pressure from a single PPG signal with a mean error of 2.95 mmHg (SD 19.33 mmHg).
Why the study?
Methods to measure continuous arterial blood pressure are often quite invasive and costly.
Can a federated learning-based generative adversarial network accurately estimate continuous arterial blood pressure from a single PPG sensor?
Can a federated learning-based generative adversarial network accurately estimate continuous arterial blood pressure from a single PPG sensor?
Effect estimate: Mean error 2.95 mmHg (SD 19.33 mmHg)
A novel federated learning GAN framework demonstrates the feasibility of estimating continuous arterial blood pressure from a single PPG sensor, potentially enabling non-invasive BP monitoring via wearables.
May enable cuffless ABP monitoring in research; leaves open prospective clinical validation before practice change.
Ischemic heart disease is the highest cause of mortality globally each year. This puts a massive strain not only on the lives of those affected, but also on the public healthcare systems. To understand the dynamics of the healthy and unhealthy heart, doctors commonly use an electrocardiogram (ECG) and blood pressure (BP) readings. These methods are often quite invasive, particularly when continuous arterial blood pressure (ABP) readings are taken, and not to mention very costly. Using machine learning methods, we develop a framework capable of inferring ABP from a single optical photoplethysmogram (PPG) sensor alone. We train our framework across distributed models and data sources to mimic a large-scale distributed collaborative learning experiment that could be implemented across low-cost wearables. Our time-series-to-time-series generative adversarial network (T2TGAN) is capable of high-quality continuous ABP generation from a PPG signal with a mean error of 2.95 mmHg and a standard deviation of 19.33 mmHg when estimating mean arterial pressure on a previously unseen, noisy, independent dataset. To our knowledge, this framework is the first example of a GAN capable of continuous ABP generation from an input PPG signal that also uses a federated learning methodology.
No takes yet. Share an insight, caveat, or question.
Brophy et al. (2021) studied this question. Time-series-to-time-series generative adversarial network (T2TGAN) was evaluated on Mean arterial pressure estimation (Mean error 2.95 mmHg (SD 19.33 mmHg)). A time-series-to-time-series generative adversarial network (T2TGAN) estimated continuous mean arterial pressure from a single PPG signal with a mean error of 2.95 mmHg (SD 19.33 mmHg).