What You’ll Learn:
- Bioimpedance can be measured using graphene-based tattoos
- This is how these readings translate into blood-pressure readings.
- This system is based on electronic tattoos and machine learning.
Although it is one of the most vital indicators for overall health and heart health, it can be difficult to continuously measure blood pressure (BP). While wrist-based smartwatches seem to offer a solution, they can’t provide consistent, accurate results because these watches slide around on the wrist and may be far from arteries.
Traditional cuff-based devices, which constrict around your arm to provide a reading, are now the standard. However, it would still be extremely useful to have continuous monitoring of pressure variations and their variations for both immediate and predictive analysis.
Now, researchers at Texas A&M University and The University of Texas at Austin have devised, built and tested (admittedly on a limited group) an approach that enables continuous non-invasive monitoring. This method provides more detailed information about blood pressure than the one-shot cuff.
The graphene-based electronic tattoos (called GETs), can be comfortably worn on the wrist for many hours. They provide continuous blood pressure measurements with an accuracy that is far superior to most other options available. Continuous monitoring of the etattoo’s blood pressure allows it to be measured in any situation, such as when there is high stress or while you are sleeping, exercising, and even during sleep. (Fig. 1).
1. Illustration of Z-BP measurement modality: (a) Three-dimensional schematic of the GETs placed onto the participant’s wrist over the radial artery, with two outer tattoos used for ac injection, and two inner tattoos used to measure voltage changes. (b) Photograph of 12 GETs, each with a surface area of 25 mm², placed on the radial (six tattoos, comprising Bio-Z1 and Bio-Z2) and ulnar (six tattoos, comprising Bio-Z3 and Bio-Z4) arteries on a participant’s wrist. For effective BP capture, multiplexed tattoo placements are essential. For visibility purposes, the arteries were pseudo-colored pink. Their locations were tracked with an ultrasound Doppler probe. Close up view of six GETs located at the radial artery. As the graphene almost disappears, the injecting GETs have been pseudo-colored with green and violet for the Bio Z1 and Bio Z2 pairs. (d) A cross-section of six GETs. The green lines represent the ac injected signal, and the gray lines the voltage sensing. (e) A close-up of one pair sensing GETs, and an equivalent simplified electrical circuit for the interface. ZTissue Zartery, part of which (ΔZarteryThis is due to an undulating volume of blood. (Credit: a, Jo Wozniak, Texas Advanced Computing Center)
“Taking infrequent blood-pressure measurements has many limitations, and it does not provide insight into exactly how our bodies are functioning,” said Dr. Roozbeh Jafari, professor of biomedical engineering, computer science and electrical engineering at Texas A&M and the co-leader of the project.
What is the Work of a Tattoo?
Injecting an electrical current through the skin to measure bioimpedance is how the device works. (Fig. 2).
2. Diagram of 13 GETs placed on the left wrist to provide BP estimation. Current injection tattoos have been colored magenta. Sensing tattoos are in light blue. The reference electrode has been colored light orange.
There’s a correlation between bioimpedance and changes in blood pressure that has to do with blood volume changes. However, as with nearly all medical-sensed data, the correlation isn’t especially direct or clear, so the team had to create a machine-learning model to analyze the connection to get accurate blood-pressure readings, in addition to the custom circuitry (Figs. 3 and Figs. 4)
3. Block diagram of the Bio-Z signal-acquisition hardware: The transmit (TX) unit injects a gain and frequency-programmable ac current at 10 kHz. The transmit (TX) unit amplifies the Bio-Z signal via IA, and then filters it through an anti-aliasing filter (LPF), with a cutoff of 30 KHz. After the filtering is complete, the signal is sent to the computer for digital signal processing (DSP).
4. The custom-made PCB designed for BioZ sensing hardware. Although it is compact, the board integrates discrete components and operates 10 concurrent channels. This allows the board to cover a variety of applications that need low-noise operation. When the circuitry is tuned to one application, it can be easily incorporated into a much smaller, and even more flexible, form-factor circuitboard.
Their self-adhesive low-impedance graphene electric tattoos will settle to the skin over time and sense from the same place. The BP estimation model used for tattoo placement can be determined right away without having to calibrate the model each time. Also, the graphene-enabled impedance BP (Z-BP) doesn’t suffer from electrode misplacement or other sensing issues.
This contrasts with wearable electrodes that are based on optical, pressure or acoustic methods. You can read the Supplementary Information post to learn more about these alternative methods.
Seven subjects were used by the team for their testing (Fig. 5) We compiled extensive data. Having a small number of subjects to test devices is a problem when it comes to testing medical applications. This is true even for non-invasive and non-risky tests.
5. Schematic illustration showing diastolic (DBP – dark Orange) and systolic (SPB – maroon), dynamic changes in different exercise routines. (a) The handgrip routine (HGCP) involves the subject performing handgrip for three minutes and slowly raising DBP and SBP. After that, the subject will place their hand inside an ice-cold bucket (coldpressor, CP), for one minute. To ensure that BP rises first, and then slowly falls over the four-minutes. Left side depicts the gradual increase in DBP and SBP, with a steady decline following the CP phase. The subject performs a four-minute set of bicycle-cycling treadmill exercises (b) while being seated. There is a four minute break between each exercise. The illustrative time traces show that cycling causes a rapid drop in BP after the subject has stopped exercising. There is also a low increase in DBP and SBP amplitudes. The Valsalva routine, which involves a subject pinching his nose and trying to exhale deeply for between 20-30 seconds, is shown in (c). This creates an intense buildup of internal pressure that causes BR to rise, decrease, and then rapidly increase again. In all of the images, the Y-axis scale has been considered equal. This is to show that HGCP and Valsalva routines have a higher change in BP amplitudes compared with cycling. Since the Valsalva routine is very quick, it’s prone to estimation errors, as the Finapres BP recordings are very noisy. A slight decrease in the original BP trace is possible by smoothing the original BP timeline.
It is the accuracy of your approach that matters most. Their data was analyzed extensively and the team presented the findings with many graphs.
A single summation number is also possible. They claim an accuracy of 0.2 ± 4.5 mm Hg for diastolic pressures and 0.2 ± 5.8 mm Hg for systolic pressures, a performance equivalent to Grade A classification for blood pressure.
(For perspective, note that most biologic phenomena are low and slow: a blood pressure reading of 100 is fairly low pressure on the broader engineering scale at just 0.133 bar, 13332 Pascals, or ~2 psi–and you’re trying to sense those low levels with accuracy and consistency despite sensing and electrical noise.)
Eight-page document describing their work “Continuous cuffless monitoring of arterial blood pressure via graphene bioimpedance tattoos” Published in Nature Nanotechnology. Although the paper is protected by a paywall it does contain open links to interesting videos. The good news? You can get the paper directly without needing to see the figures. Here while the captions and figures can both be found in this article, they are also available as separate documents. here (scroll down).
Finally, there’s a 51-page Supplementary information posting that has extensive and fascinating discussion of the technique, setup, comparative analysis with other approaches, extensive data, numerous photos, electronics details, signal processing, machine learning, data and error analysis (do you know what a “violin plot” of data is? You can find more information here.