Youngju Nick
The iiPillow product
displayed for demonstration
with its producers.
iiPillow: the intelligent inflatable Pillow
Project team member(s): Youngju Kim and Nick Hafer
Introduction
Sleep apnea is a near-ubiquitous problem that affects people the world over, regardless of living conditions or social status. Many of the products on the market today (e.g. CPAP machines and high-tension chinstraps) are uncomfortable or highly involved. With our new product, we aim to help consumers with sleep apnea via an elegant solution: a pillow. This isn’t just your everyday cushion, however; it's a smart pillow, featuring various pressure chambers inside that can inflate and deflate depending on the person’s breathing. Information on the state of the sleeper of will be acquired through the use of accelerometers, possibly on the body of the person: when they sense a lack of breathing motion, they’ll relay information to the pillow which will inflate, tilting the person’s head to open up their airway.
Background
As stated above, the status-quo treatments for sleep apnea include straps and headgear that help keep the airway open (ostensibly in milder cases of OSA), and the notorious CPAP (Continuous Positive Air Pressure) machines.
The primary drawback to both is comfort. The straps aren't wholly uncomfortable but are awkward enough to hinder a deep sleep, which is often what sleep apnea patients are after when they seek treatment. CPAP machines, by contrast, are uncomfortable because they keep the airway open quite literally by forcing air down the sleeping patient's throat: the positive pressure thus provided effectively inflates the windpipe and prevents the soft tissue from collapsing in and stopping the patient's breath. This isn't at all comfortable to sleepers, however, and the constant airflow can dry out and irritate the trachea.
Our pillow isn't the only attempt at an intelligent cushion that helps combat snoring (more on this in the Checkpoints below); products exist that use microphones to pick up the sound of snoring, then either gently shift the head or generate a vibration to wake the user up. However, both of these designs have plenty of room for improvement.
Construction
Pillow:
We began with a hand-me-down pillowcase that effectively constrained our maximum dimensions for the pillow, then proceeded to construct the inflatable air cells that comprise the supporting structure of the pillow. The first cell we created was a longer unit that took up the bottom half of the pillow case (to rest under your neck); we'll call it the comfort cell. The plastic tubing was heat-sealed to prevent leakage and cut to length. We then added another cell that was about ⅗ the size of the first, longer cell. This second cell acts as the main motion-driven mechanism in our project; It would (in theory) receive directions from the accelerometer and the code in order to raise or lower the back of the head and open up the airway. In order to address the possibility that the head might fall sideways off of this lifting cell, we added two square cells — completely sealed off — on either side of the lifting cell. We only had to hook up air tubes to the lifting and comfort cells, and we rarely had to inflate or deflate the comfort cell at all; this is a vast improvement from some of the more ridiculously complex designs we considered. In the context of our prototype of the pillow (which uses manual pneumatics rather than an automated pump), we also figured out that by using a four-way directional pneumatic joint, we could open up a valve in the lifting cell and it would deflate at the perfect rate out of the two open sides of the joint. This allowed us to achieve a relatively quick, yet silent and unnoticeable deflation when attempting to open the airway. Once we had all of these cells, we used a mixture of taping and stapling to connect them together. We then put our pillowcase over and it was ready to begin testing!
Attach:iiPillow2
The iiPillow without
a pillowcase, displaying
the different air cells.
Approximate dimensions of each cell:
Comfort Cell: 60 cm x 12 cm
Lifting Cell: 36 cm x 12 cm
Side Square Cells: 12 x 12 cm
Pneumatics:
We attached one tube to the comfort cell and closed the valve on it so the pressure wouldn’t change, but so that we could adjust it if we needed to; another tube connected the lifting cell to a four-way directional joint. One side of this joint was connected to a pressure gauge which led to the wall supply of compressed air, and the another side was connected to the lifting cell. The other two sides were connected to valves that we could open when we wanted to let air out.
Accelerometer:
We taped our simple 3-axial accelerometer (brand unknown) to a small square of cardboard so that it would stay facing upward in the same direction, allowing us to measure the same value along the same axis every time we looked at our code. The accelerometer was then hooked up to a modified Arduino (i.e. Hapkit Board), which plugged into a laptop to process the data coming from the sensor.
Major Component Prices:
Accelerometer: $15
Arduino Uno REV3: $22
Hapkit Board (used but not required): $35
Results
The inflatable mechanism of the pillow functioned surprisingly well! When we had people live-test the pillow's slow deflation, most agreed that their airway felt at least marginally more open, and that the gradual actuation was definitely not something that would wake up even a light sleeper.
The sensing mechanism via accelerometer was also successful, though moderately less so. The data that came from the sensor was very noisy (not a huge surprise considering the simplicity of our code), but we could still make out trends in the data and distinguish steady breathing from a cessation of the breath. Another issue was that of individual calibration; in order for our program to work properly, we had to set a specific normal value from which the program would take its measurements — the amount of tinkering needed to get the sensor to give consistent data from person to person really wasn't practical. Ideally, in the future we could clean up the code and generate a mechanism by which the program calculated change in motion consistently, regardless of the patient, and decide based on a threshold number of occurrences of the "static" position whether or not the patient was breathing.
An unexpected issue arose in the form of pneumatics: we had some issues with changing tube size (from small to large) due to the lack of adapters in the lab, and ended up resorting to double sided tape and manual pressure. Sometimes our tubing would also come out of the pillow itself, and we would have to retape that too. This isn't an issue that would be really any problem in an actual manufacturing situation, but it's worth taking note of.
In terms of improvement? There's a lot to be done. The most obvious problem with our prototype at the moment is the manual control of the air valves; in an ideal situation, the CPU would process a cessation in breathing and then send the signal to an automated air pump, which could inflate or deflate the variable cell of the pillow accordingly. Even this method has problems too, though; air pumps tend to be noisy and difficult to muffle, and waking up someone suffering from OSA with the loud whirr of an air fan is the last thing we want to achieve with our product. A possible solution comes in the form of a radically different pillow design, suggested by one of the people who listened to our presentation/demonstration: a foam cylinder mounted at a slight offset on a small electric motor — visualize a cam shaft — that would slightly tilt the patient's head up or down depending on the actuation of the motor. This would eliminate the need for a noisy pump, and could ostensibly work just as well with the accelerometer sensor we used for our prototype; how comfortable it would be is a question that's still up in the air.
In any case, regardless of the flaws in our design, we've hopefully done some productive thinking about a comfortable, easily produced treatment for sleep apnea that could eliminate some of the patient complaints associated with other treatment devices, ultimately making nights longer and easier for those in OSA.
Acknowledgments
Special thanks to Allison Okamura for helping us develop the code to interpret accelerometer readings! Another special thanks goes out to Laura Blumenschein as well for helping us set up our pneumatics system on Demo Day.
Files
Drawings, code, and anything else that cannot be directly shown in the report should be linked here. You can upload these using the Attach command.public site, please discuss with Allison. Also, in this section include a link to a file with a list of major components and their approximate costs.
- The code we used to parse the accelerometer data:
- File listing major components with approximate costs (this is also documented above):
References
Patient education: Sleep apnea in adults (The Basics): https://www.uptodate.com/contents/sleep-apnea-in-adults-the-basics?csi=9986866d-637b-46a4-a5c2-c70ca02e9733&source=contentShare
Patient education: Sleep apnea in adults (Beyond The Basics): https://www.uptodate.com/contents/sleep-apnea-in-adults-beyond-the-basics?csi=83260310-2b8a-4e66-9ecf-8b2b67696d4b&source=contentShare
Flexible Multifunctional Sensors for Wearable and Robotic Applications: Xie, Mengying, et al. “Flexible Multifunctional Sensors for Wearable and Robotic Applications.” Wiley Online Library, John Wiley & Sons, Ltd, 4 Jan. 2019, https://onlinelibrary.wiley.com/doi/abs/10.1002/admt.201800626.
ZEEQ Website: https://remfit.com/pages/zeeq
Sutherland, Kate, et al. “Oral Appliance Treatment for Obstructive Sleep Apnea: an Update.” Journal of Clinical Sleep Medicine : JCSM : Official Publication of the American Academy of Sleep Medicine, American Academy of Sleep Medicine, 15 Feb. 2014, www.ncbi.nlm.nih.gov/pmc/articles/PMC3899326/.
<<<<<<<
>>>>>>>
Checkpoint 1
Checkpoint 1 goal: become familiar with the symptoms of sleep apnea and whether our pillow can effectively treat them, and construct a design rationale based on said symptoms.
We’ve done some background research on sleep apnea to confirm whether or not the directional approach we envisioned for iiPillow is actually viable, or if we need a different technique. We also scoped out the playing field in terms of similar products already on the market, and discovered the Smart Nora pillow, which uses an audio-based sensor to detect when a person starts snoring/stops breathing and gently inflates the pillow to bump the person's head up and open up their airways.
This kind of inflation as opposed to directional design is a possible component to consider in our prototyping process; it could certainly be a much simpler mechanism to work with, as it'd require less inflatable compartments. It's also worth noting that though the audio sensing idea is interesting, it only works for people whose homes are in relatively quiet environments with little to no noise pollution; by using accelerometers as we're currently planning, we could still create a novel product that works even for those in urban, louder areas.
All said and done, we've understood much more about sleep apnea and have also come to some potential conclusions about the design of the pillow.
Checkpoint 2
CHECKPOINT 2: construct the pillow prototype and make final decisions on the design of the prototype sensor
Upon revision of our initial design (composed of a pair of 6x6 "packs" of inflatable plastic heat-sealed cells), we realized that it was needlessly complex given that the only actuation necessary to help open up the airways was a gentle tilting back of the head; we therefore simplified our design to one constantly inflated insert at the front and a single cell situated at the center rear of the pillow that slowly deflates to allow the head to rock back. We also eliminated any extra padding for the moment, as the inflatable cells alone turned out to be surprisingly comfortable already.
The issue of air cable management was hence mitigated fairly simply, as we now only have two air cables running from the pillow (one for the largest front cell — which can be inflated or deflated to better suit the head size/shape of the user — and one for the smaller variable cell situated in the back). They're currently inflated/deflated using manual control of the valves, as developing an automated control system for our iiPillow seems out of the scope of this project — for the moment, at least.
That being said, we still need to integrate a viable sensor with which to measure breathing into the product. We've finalized the decision to use an accelerometer-based sensor that will track movements of the chest and/or stomach and (ideally) distinguish between the regular rise-and-fall of a comfortable sleeper and the aggravated snoring then cessation of breath characteristic of people suffering from OSA. We have yet to manage this and are hoping to arrive at a working prototype of the sensor within the next two days.
So that's where we are!
Cheers, iiPillow development team