To make a Goldilocks face-mask: It complains if worn too loose (transmitting droplets) or too tight (asphyxiating), happy when worn 'just right'!
Executive SummarySince the start of this Covid-19 crisis, the humble mask has been dragged into the spotlight, reviled and celebrated depending upon, strangely enough political and cultural preferences. This has not been helped by the muddied messaging adopted by political leaders in the 'mature western' democracies. This becomes even more alarming when we consider that leaders in developing countries, where resources to manage pandemics are limited, often take their cues in handling global crises from their counterparts in western democracies.
On the other hand, the medical community and the World Health Organisation, is steadfast in its recommendation of the wearing of masks in public to limit the spread of the virus. This is predicated on the public adopting the measure wholeheartedly, recognising that if we all wear masks we protect each other. The question this project aims to answer is - how can scientists steer the messaging around masks so that objective facts rise above subjective opinion.
This project focuses on an often times cited yet sometimes valid concern, i.e. a sense of claustrophobia. The perception of suffocation may arise either when first putting on a mask, or due to evolving conditions whilst wearing a mask, as stress inducing conditions arise (such as running). Initial discomfort may deteriorate to a a very real onset of asphyxia, or choking, due to deficiency of oxygen supply to the body. This onset of oxygen deprivation in the blood stream, can get worse if the individual panics leading to hyperventilation. In monitoring the SPO2 levels in real time as the wearer goes about his or her daily business, the user (or a carer) can proactively alter the tension in the mask to avoid conditions that lead to suffocation. The innovative step in this project is an SPO2 sensor instrumented face mask streaming data in real-time to an online browser accessible dashboard.
MotivationThe personal motivation to embark on this project is a complex web of reasons: part personal and part scientific.
The personal element dates almost six years back when I noticed my father in his last few months in his battle against cancer, was often overcome by a sense of suffocation coupled with dropping SPO2 levels when, paradoxically, fitted with an oxygen mask. At the time I attributed the condition to his body's deteriorating state. Cut to present day, a few weeks ago, on our periodic walks around the picturesque North Yorkshire village we live in, I could see my elderly (albeit reasonably fit) mother visibly struggling to breathe when climbing uphill wearing an N95 mask.
Reading more into the published advice around face masks, one learns that the feeling of suffocation is not necessarily a perception but could be a very real condition when the wearer of a face mask, undergoes further stresses - either of an external nature (walking uphill!) or internal (low lung capacity), which can in turn lower the oxygen levels in the blood. Although the bulk of the population that is young and healthy with no underlying conditions will be perfectly fine going about their daily duties wearing a mask, one cannot rule out a subset of the population being adversely affected. This then begs the question: can face masks be made safe for all? Only then can they be made mandatory for all.
My scientific motivators are a combination of exasperation and curiosity. My exasperation lies in the fact that the mere wearing of a face mask seems to have become a political symbol, a clear casualty of these polarized times. The frustration is that the majority of the scientific community, which hold facts above all else, is now being dragged into this warped reality by a small group of vested interests. And my curiosity was triggered on reading more about the requirements and recommendations behind face coverings and how something like thread count impacts breathability as well as droplet transmission (Ref [1]). Digging a bit more through interviews (Ref[2]) as well as advisories around medical advice 'for the-non-medical-person' (Ref[3]) led to the oxygen saturation metric used to track blood oxygen levels. A recent interview citing a medical doctor wearing a face mask and a pulse oximeter to measure oxygen saturation or SPO2 effectively clinched it (Ref[4]). We all need an ideal face-mask: One that tells us if its too loose (letting droplets out) or too tight (choking). Heck we need a Goldilocks face mask!
Bringing it back to the personal: I identify as a scientist first. A scientist I like to think, looks for the rational to explain the seemingly irrational. And in an increasingly hysterical world, I believe it is up to the scientific community to bring objective reasoning back into the public conversation. I also believe that the majority of people don’t really care for the existing political polarization or indeed have an opinion on it. They know that they don’t know enough about the science, are bewildered by the mixed political messaging, yet they want to do right by their families and society. As scientists and engineers I strongly believe this COVID19 Detect and Protect Challenge is fantastically poised in helping precisely those people, and I would like to think that this MaskSpoT project could help move the needle in the right direction.
ConstructionTo understand the path towards building the MaskSpoT sensor here we outline the essential requirements, the initial ideas tried and discarded and finally an outline of the actual sensor itself.
Basic requirements:
- Am I choking: Typical pulse oximetry indications suggest anything below an SPO2 of 95% is a source of concern (Ref[5]). The requirement is to monitor the SPO2 levels and advise on dangerous levels.
- Is it on: Sensors in the MAX30105 allow us to track proximity to the wearers face. We need to exploit this to track when the mask is not being worn.
- Minimal fuss: Wearing an instrumented mask should be no more complicated than wearing a normal mask!
- Real time information: Data streamed to a browser (or an app_ allows for real time monitoring and immediate decision making.
- Simplicity: Design and construction, for the wearer as well as the manufacturer!
Initial prototyping and ideas
- Proximity & Temp and Humidity: The initial prototypes tested (Ref sketch1 ) examined options of placing proximity sensors near the cheek and temperature and humidity sensor (AM2320) in the mask. Although these worked on the bench-top the complexity of design belied the usage. Nonetheless the proximity tests showed IR sensors work across a broad range of raw values which could be exploited to identify the mask state (on or off). The SPO2 estimation proved to be more interesting and this early prototype was abandoned in favour of final design.
- Specific Blood Oxygen: The chosen sensing protocol adopted was to harness the blood oximetry measurements made possible by the MAX30105. The sensor is a reflectance mode oximetry type. Note that two common types of pulse oximetry sensor configurations, are transmission mode and reflectance mode.
- Transmission mode: the optical emitter and detector are positioned on opposing surfaces of a body extremity (such as a finger as seen in pulse oximetry sensors).
- Reflectance mode: the emitter and the detector are adjacent to each other, and are applied to one surface of the body, such as the cheek as employed here.
- In either modes, the emitter shines red and infrared light into the skin and the detector measures the scattered light that is transmitted through blood-perfused tissues. Reflectance mode sensors have been tested in applications involving the forehead (Ref[6]) and cheek (Ref [7]), with tests indicating that sensors reflecting off the cheek respond faster to changes in blood oxygen levels. This is useful for our configuration and has hence been deployed here.
Final Design
: Proximity & SPO2 with
the MAX30105
The full layout of the mask and attached sensor with microcontroller is shown In Sketch[3]:
- The mask is fitted with MAX30105 with a slot cut-out for the sensors to rest in contact with the cheek of the wearer.
- As indicated in the circuit diagram, the sensor is connected to the Arduino Feather Huzzah, which is powered by a 3V battery pack.
- In the current configuration the feather and battery pack are worn by the user on their person - this is currently slipped into a leather sleeve that can be fitted onto the wearers clothing (much like one would wear an Apple iPod Nano when running). For the current design the 4 single core wires ensure sufficient capacity to transfer voltage and sensing signals, with some give to minimize tension on either soldered ends.
- The final part of the sketch shows the realtime MaskSpoT Dashboard on Adafruit IO from the two feeds sent periodically by the Feather: maskon and spo2
Under the hood the system works as follows:
- The MAX30105 generates estimates for SPO2 levels using the red and IR sensors The proximity sensor calculation is run parallely. These two measurements allow us to estimate the two questions we seek to answer with this project:
- Is the mask too loose: Proximity sensor. The IR sensor takes on a set of raw values, the range being determined based on the reflected light detected by the particle sensor. Based on empirical testing we hardwire the code to select a mid-range value that acts as a cut off to indicate if the sensor is too far from the skin and triggers the mask-off state. Based on the threshold cutoff selected, the Feather sends a 1 or a 0 to the maskonFeed.
- Am I choking: SPO2 sensing : For spo2 level calculations we deploy the logic in the code published by Sparkfun with the MAX30105 sensor. SPO2 calculations are derived from Red and IR led reflectance measurements. This spo2 level is computed as a percentage and is sent to the spo2Feed.
- The Feather requires wi-fi connectivity to transmit data to the Adafruit IO feeds. For outdoor operations, the device can be connected to the wearers mobile phone configured as a Mobile Hotspot.
- The settings on the MaskSpoT Dashboard are configured based on published medical advice (Ref[5]) and are as follows:
- Mask too loose: maskon Feed
- 1: Mask on, setting is green
- 0: Mask off, setting is red
- Mask too tight: spo2 Feed
- 95-100: Gauge level, indicates normal levels, colour Blue
- 80-95: Gauge level is low, indicates warning, colour amber
- <80: Gauge level is below minimum, indicates danger, colour red
The design although promising is far from the finished product. Some aspects need further consideration and experimentation in order to find their way into a finished product. A few items that made it on to the list are:
- The flexible tinned copper wires is the principal aspect that requires more formal consideration. This modification should be reasonably straightforward to replace with an off the shelf 4-cored wire.
- The Feather Huzzah can be replaced with a Feather FONA. This would allow outdoor data streaming operations without the need for a supporting phone acting as Wifi server. the Feather Bluefruit LE is another efficient low power option, which would however need bluetooth support from a mobile phone
- The sensor has been tested on reasonably healthy subjects. There are additional aspects that need consideration -
- SPO2 levels: Need to be calibrated with an in-hospital monitor on a wide range of people, age groups, COVID-19 co-morbidities and COVID19 survivors.
- Dashboard configurations: In these COVID times 'normal' levels of SPO2 might be individual specific and could vary considerably with lung condition (Ref[5]). Do we need a mask with an oximeter settings that is designed to be configured to an individual's body (probably)? How can we enable this at first user setup? Is a self calibration system needed, and if so possible?
- Reflectance oximetry: the sensitivity to ambient light conditions needs to be better understood
- Feather Huzzah : if used, we need to develop an app that enables the user to configure wifi settings.
- Size and form factor optimization: this configuration involves a support casing with battery worn on the users person, to avoid the mask becoming too heavy. It is likely a smaller, slimmer form factor design can be identified to ensure the mask is self contained?
- Environment proofing - Rain and heat are two environmental conditions that will adversely affect the instrumented mask. Although some silicone protection could be deployed for water proofing, battery operations under hot temperature conditions are a concern.
Ref[1]: How to select a mask: https://fivethirtyeight.com/features/what-to-look-for-in-a-face-mask-according-to-science/
Ref[2]: Wearing Masks: https://fivethirtyeight.com/features/the-science-of-mask-wearing-hasnt-changed-so-why-have-our-expectations/
Ref[3]: Killer Covid-19 Masks: https://hartfordhealthcare.org/about-us/news-press/news-detail?articleid=26712&publicId=395
Ref[4]: Mask with Pulse Oximetry
Ref[5]:Is My Blood Oxygen Level Normal?: https://www.healthline.com/health/normal-blood-oxygen-level
Ref[6]: Reading SPo2 on Forehead: https://anesthesiology.pubs.asahq.org/article.aspx?articleid=1931026
Ref[7]: Reading Cheek data: https://pubmed.ncbi.nlm.nih.gov/8466011/
Comments