Embedded Files
The Challenge
Cardiac arrest is one of the leading causes of death worldwide, with over 17 million deaths each year. In these emergency scenarios, quick intervention can often be the difference between life and death.
A common issue in these emergencies is being able to identify a pulse, and being able to discriminate between a live and non-live one - in other words, being able to clearly distinct between a beating heart and a non-beating heart.
In today's world, manually measuring the pulse (with fingers) is the often the most reliable and quick method for establishing the presence of a heartbeat. However, this can be challenging, taking up to as long as 10 seconds to recognize a stable pulse. Moreover, a manual pulse check while performing CPR is essentially impossible for a single physician, prompting the need for more medical staff.
Other practices can rapidly deliver accurate pulse detection, but follow invasive methods that require time to prepare, making it impractical for emergency situations.
With a team of Electrical engineering students of diverse experiences, we set out to innovate a non-invasive, hands-free, portable, rechargeable, compact and reliable pulse-detection device that can accurately distinguish live and non-live pulses - known as LifePulse.
Technology
The device uses photoplethysmography (PPG) technology, a non-invasive optical technique that measures the changes in blood volume.
Similarly seen in Apple Watches, PPG technology uses light-emitting diodes that project light rays into the blood. Oxygenated blood cells absorb the light, and all other light is reflected back into photodetectors that measure its amount. As your heart beats, blood is moved continuously through the body in which the photodetector will measure its variance, and from that, we can calculate a heartbeat value.
PPG technology can also be seen in pulse-oximeters - devices clipped onto small parts of the body such as the finger or ear. As opposed to smart watches, pulse-oximeters will measure the changes in transmitted light.
From the Outside
The pulse detector is built in a compact and portable device with sizing comparable to wireless earbud casings.
The device can be turned on-off with a simple switch on the side, operational capability for over 30 minutes and fully rechargeable via micro-USB in less than 15 minutes. Hands-free pulse-detection, with visual and auditory features for notification.
The bottom of the device is arched to adhere to the natural neck curvature, and is where the sensor protrudes to be close to the skin. This acts as a patch-like, hands-free device so physicians can receive a pulse reading while performing other important tasks.
Design
The device comes together with a number of electrical and mechanical components, visible in the diagram to the left. This assembly consists of 10 individual pieces.
ELECTRICAL
The heart of the device's functions is in the ESP32 microcontroller, which is merged to exist as one piece with the liquid crystal display (LCD). In this microcontroller is programmed an algorithm to boot the device, initiate detection, and process incoming and outgoing signals.
The responsibility of detection lies within the created breakout board (in green), which holds a PPG medical-grade sensor that carries four different LEDs and a photodiode. The board picks up the detected signals and transmits them to the microcontroller.
The remaining electrical components control the rest of the device's functionality - power board for battery management and rechargeability, battery for power, speaker for audio, and a switch to boot the device.
MECHANICAL
The mechanical enclosure is composed of four different 3D-printed parts, with each fitted to its neighboring piece. The parts are visible in the exploded-view, with one being semi-transparent for display purposes. In each piece carries M.2 screw holes and is printed using medical-grade filament.
The base is composed of two pieces, in which they act as the foundation to carry everything that follows above. An opening is made on the side for the switch to extrude out of the enclosure, making the on/off operation accessible.
The middle piece is referred to as the LCD-Base, in which the microcontroller + LCD is placed upon. This piece is responsible for keeping the microcontroller and the power board tightly secured via screws, and encapsulates the remaining components.
The lid acts as a barrier for the liquid display so that users can use the device without concerns of being in direct contact with the screen.
Top Figure: Illustration of process in filtering rising edges (2)
Bottom Figure: Comparing remaining rising edges for similarities to validate pusle (3)
Signal Processing
The signal processing utilizes an implementation of the Nenova Algorithm.
The Nenova Algorithm is an automated algorithm designed for false pulse wave detection, useful in cases of emergency cardiac activity. It is essentially a multi-step sequential process, which after preprocessing of small ripples and variations (with a moving average filter), follows:
Finding maxima and minima based on signal segments
Filtering rising edges using pairs of peaks and troughs
Validating pulse based on remaining rising edges
Calculating heartrate with distance between adjacent maxima
The Nenova Algorithm was chosen because of its capability to achieve certain performance metrics - > 90% sensitivity, > 70% specificity, down to a 4-second detection time, and < 15% heartate error. While it is one of the more complex algorithms having many tunable parameters, for our objectives the pros heavily outweigh the cons.
Summary of Results
After testing, the device successfully passed all validation requirements - tested for functionality, usability, and power.
The verification was performed based on two significant aspects - pulse detection accuracy and heartrate measurement accuracy. One to verify the device can differentiate between a true and false pulse, and one to verify the correctness of the heartrate measurement. These tests were done simultaneously with either an Apple Watch or a pulse oximeter.
From the graphs illustrated, we can see that in the first figure, our device achieved a pulse detection accuracy over 71% sensitivity and 95% specificity, which beats metrics of manual palpation at 70% sensitivity and 90% specificity.
Sensitivity and specificity differ in that one measures the performance on the ability to detect a positive, and one measures the performance on the ability to detect a negative. In our case - sensitivity measures the ability to detect true pulses, specificity measures the ability the detect false pulses. We can also refer to the formulae to mathematically describe the difference:
The second figure depicts heartrate errors, in which 91% of values taken are within 15% of their true values. We use this 15% bound as most heartrate measurements can have up to 15% of error.
Illustrative chart of collected data in TPs, TNs, FPs, FNs
Sample Count vs. Percentage Error in collected data
LifePulse - Version 2
While Version 1 successfully delivers a pulse-detection device that fulfills our requirements, there was still a growing urge to compact the device even further. As it stands, Version 1 is still slightly bulky - numerous components and a relatively heavy microcontroller unit - which can interfere with its ability and duration to stay adhered to a patient's skin.
With this in mind, the goal for Version 2 was to downsize the device - primarily by reducing the number of components needed to create a functioning device.
In combined efforts, every electrical component (besides the battery) was combined into a compact circuit board. This automatically downsized the mechanical enclosure, making the device much more compact - including a 50% reduction in height, 16% reduction in length, and 5% reduction in width.
What I learned
LifePulse was an extremely insightful project that gave me perspective on developing an entire product from start-to-finish. The design shed light on several aspects of work - firmware, software, hardware - which were all vital to completion of the device. This project has also helped me better learn the environment of working in a tightly knit team and how to collaborate effectively to use the best of each individual's differing experiences. Most notably however, the project introduced me to development in the medical field and with it, the experience of designing a device with intent to directly save lives.
Page updated
Google Sites
Report abuse