3D-Printed Finger Clip Sensor, Linked to an ML System, Offers Five-Second Blood Pressure Readings
Designed for ease of use and speed of reading, the machine learning-backed fingertip sensor is 90 percent accurate for systolic readings.
Engineers at the University of Missouri have created a 3D-printed prototype blood pressure monitoring device, which only needs to be clipped to a finger, like a blood-oxygen monitor — offering accurate readings within five seconds.
"Typically, calculating someone’s blood pressure at a hospital or clinic involves using an inflatable cuff wrapped around their arm, but there are three issues with that method," explains Richard Byfield, lead author of the study. "It can cause damage to someone’s arteries if done repeatedly within a short amount of time; people’s blood pressure can rise due to nervousness; and it can take up to 30 seconds to complete. Our device can record someone's blood pressure within five seconds by using optical sensors placed on the fingertip that measure the amount of light reflected off the blood vessels underneath the surface of the skin."
Based on customizing a commercial finger clip device, of the sort typically used for measuring blood oxygen levels, the prototype device features two photoplethysmography (PPG) sensors to read pulse wave velocity - how rapidly blood is travelling through the bloodstream — at two points on the finger.
These two readings, which on their own say little about blood pressure, are transmitted wirelessly to a host computer which processes them through a machine learning algorithm capable of inferring blood pressure based not only on each individual reading but also on the difference between the two — offering, the researchers found, a 90 percent accuracy rate for systolic blood pressure dropping to 63 percent for diastolic.
"Typically, there are a few problems with PPG sensors," Byfield explains of the issues his team had to resolve. "One is called artifact motion — if you move a PPG sensor while it’s reading, it can affect the waves that are being recorded.
"On top of that, we found that differences in pressure can alter the waves, but with a finger clip design, a spring provides constant pressure. Another reason this method hasn't been explored much before is typically these finger clips only have one sensor, but we have two sensors in our device."
The team is hopeful the device would be suitable for at-home use, and could provide data for studies on using vital signs for indicators of a range of diseases — including COVID-19. "Our goal is to develop a broader impact for our device," says Jian Lin, associate professor and corresponding author, "beyond a new way to measure vital signs."
The team's work has been published as easy access in the journal IEEE Sensors, under closed-access terms.
Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire: freelance@halfacree.co.uk.