Researchers at Johns Hopkins University have developed a system that uses wearable sensors and AI to measure blood pressure continuously with almost the same accuracy as an invasive arterial catheter. Initial trials in intensive care units show that the technology could become a less invasive alternative for continuous blood pressure monitoring. In the long term, the researchers also see potential applications outside the ICU and for home monitoring in people with hypertension.
In intensive care units, continuous blood pressure monitoring is vital because the blood pressure of critically ill patients can fluctuate significantly. High blood pressure increases the risk of strokes, heart attacks and kidney damage, amongst other things. If blood pressure is too low, the brain and other vital organs may not receive sufficient blood.
An alternative to the arterial line
For continuous monitoring, an arterial line is currently often used: a catheter inserted directly into an artery, for example in the arm or groin. This method provides accurate real-time information, but carries risks of bleeding, blood clots and infections. Furthermore, an arterial line restricts patients’ freedom of movement.
Whilst a traditional blood pressure cuff is non-invasive, it only measures blood pressure at specific intervals. The researchers therefore sought to combine the advantages of both methods: continuous monitoring without a catheter. The system they developed, MOSAIC, uses two wearable sensors. One sensor is placed on the chest and the other on a finger. Together, they record the heart’s electrical activity and blood flow throughout the body. These signals are processed by a deep-learning model.
The AI uses this data to reconstruct a so-called waveform: a continuous curve of arterial blood pressure. According to the researchers, this makes it possible to obtain relevant and reliable blood pressure information without an invasive procedure.
Initial trial on ICU patients
MOSAIC was initially tested on 28 patients in the intensive care unit at Johns Hopkins Hospital. According to the researchers, the blood pressure curves generated by the AI model closely matched the measurements taken by traditional arterial catheters. The research team is now working on validating the system with a larger group of ICU patients. This should clarify whether the technology is sufficiently reliable and robust for wider clinical application.
If further development proves successful, the system could also be used outside the intensive care unit. Continuous blood pressure monitoring could, for example, become possible on standard wards, where patients are not usually monitored via an arterial line. In addition, the researchers see potential for home monitoring of people with hypertension. Wearable sensors could provide insight into changes and trends in blood pressure over longer periods during everyday life. The researchers compare this potential development to continuous glucose monitoring in diabetes.
Even in healthy people, long-term monitoring could yield new information about how blood pressure changes during normal daily activities. As a result, the technology could ultimately not only offer an alternative to invasive monitoring but also enhance our understanding of blood pressure outside the hospital setting.
Smartwatch
Earlier this year, we wrote about the development of a smartwatch capable of continuously measuring a person’s blood pressure without the need for a cuff. The technology uses an imperceptible electrical current to measure bioimpedance. Changes in this provide information about blood pressure and blood circulation.
The measurement data is processed by an AI model that integrates physical principles from fluid dynamics and electromagnetism. The researchers aim to use this to improve reliability and interpretability compared to AI models that primarily function as ‘black boxes’. Instead of recording only systolic and diastolic values, the system records the full blood pressure waveform. This allows changes during daily activities to be monitored continuously. Furthermore, individual calibration is not required. The technology was tested on approximately 150 people, including ICU patients and outpatients.
References
Computers in Biology and Medicine (research)
