Physics-based machine learning enables continuous blood pressure monitoring without a cuff

Blood pressure is an important measure of cardiovascular health, but the standard method of measuring blood pressure relies on occasional measurements, usually in a clinical setting using an inflatable cuff. Current blood pressure monitors are large and uncomfortable to use, and can only provide readings when you are sitting still.

Now, an interdisciplinary team of mathematicians and engineers from the University of Utah and the University of Illinois at Chicago is tackling this challenge by combining physics and artificial intelligence to overcome some of the limitations of existing devices. Coming soon nature communicationstheir study describes a new wearable smartwatch that can continuously measure both blood pressure and blood flow without the need for a cuff.

“Elevated blood pressure is considered a silent killer that causes heart attacks, aneurysms, and strokes. Elevated blood pressure is a global health burden and is considered a holy grail problem,” said Benjamin Sánchez Terrones, who initiated the project several years ago as an assistant professor of electrical and computer engineering at Utah State. It works by measuring the electrical properties of blood passing through arteries in your wrist. The electrical properties of blood change in response to changes in blood pressure.

The University of Utah owns the intellectual property related to this technology, which is based on physics-based machine learning, and the university’s Technology Licensing Office is currently exploring licensing opportunities to bring this invention to market.

light and electricity

The scientific basis for commercially available wearable devices that use light to estimate blood pressure is not fully understood, and they often rely on machine learning as a “black box” to determine blood pressure, making their output difficult to interpret and clinically trust, the latter being a major barrier to clinical adoption. Unlike these devices, which measure blood pressure by measuring light, Sánchez-Teronez’s device uses a painless, imperceptible electric current.

This technology records small electrical changes in the wrist using bioimpedance, which measures how easily electricity flows through blood and tissues. As blood flow changes with each heartbeat, these electrical signals convey information about the underlying pressure.

This research shows that combining machine learning and physics can fundamentally change what is possible. By incorporating physical principles directly into models, we can move beyond black-box predictions toward systems that are more accurate, more interpretable, and broadly applicable to real-world medicine. ”

Christel Hoenegger co-author, Associate Professor of Mathematics

The role of fluid mechanics and electromagnetism

The system uses hydrodynamics (blood flow) and electromagnetism and has a clear scientific basis, making it more reliable. This model encodes the physics of pulsating blood and the electromagnetics of bioimpedance measurements, so the network does not predict anything that is physically impossible.

The result is a wearable device that can continuously track cardiovascular health during rest and activity, without the need for individualized adjustments.

Utah graduate students Henry Crandall, Tyler Schuessler and Philip Belik played a key role in testing the device on 150 real-world patients, including those in intensive care units and outpatient settings. “We went a step further and measured patients in the intensive care unit and Madsen Health Center. [a clinic just off campus in Salt Lake City] Because we wanted to test this technology on a targeted population,” Sánchez Terrones said. Last year, he moved his lab to the University of Illinois at Chicago, where he is an associate professor of electrical and computer engineering and biomedical engineering.

blood and movies are similar

“Your blood pressure throughout the day is like a movie, but when you put the cuff on, all you get is one snapshot of the picture,” says Sánchez-Teronez. “While cuff devices are extremely useful, they also have their limitations. The way the technology works, systolic and diastolic measurements overlap and translate into maximum and minimum pressure values ​​during the recording, giving minimal useful information. Finally, 99% of the film is missing, explaining how blood pressure changes as the patient walks, runs, and climbs stairs throughout the day.”

Sanchez Terrones’ technology can capture the rest of the movie by recording blood velocity and pulse as a continuous waveform, as well as the well-known systolic and diastolic values ​​provided by standard cuff measurements such as 120/80. (Systolic is the top number that measures the pressure on the artery walls when the heart contracts, and diastolic is the pressure when the heart is resting between beats.)

“Blood pressure is a function of time, not two numbers. The mathematical challenge was to recover that entire waveform from indirect electrical measurements at the wrist, a classic inverse problem,” said co-author Braxton Osting, a professor of mathematics at the University of Utah. “Embedding the physics of blood flow directly into the model increases the confidence in our predictions.”