There’s Coronavirus in the air. Can it be detected in real time?

If you’re the Ontario government, you answered yes. In fact, the provincial government was so impressed by a specific application of this technology that they committed $2 million in funding. There’s just one problem. Well, actually, there are many.

CEM Specialties Inc.’s (CEMSI – based in London, ON) Kontrol BioCloud is a device its inventor describes as a “game changer” in the fight against the SARS CoV-2 virus. The device was designed to be installed in a room where it will continuously monitor the air for traces of virus. Upon detection, the system sends alerts to those designated to receive them, empowering them to make decisions based on the presence of the virus within their facility. The device’s promotional material asserts that these decisions will help mitigate outbreaks and speed up contact tracing.

CEM Specialties Inc’s president Gary Saunders has said this technology will further help diminish the spread of the virus. Across a number of press releases and interviews, the company and its representatives have floated potential clients for these devices: schools, hospitals, factories, transit systems, etc. Essentially, every shared space where there is concern of COVID transmission is a business opportunity. According to Kontrol Energy Corp. (parent company) CEO Paul Ghezzi, “BioCloud will not just monitor air quality, it seeks to create the conditions for Canadians to safely and confidently return to their jobs and their schools.”

This all sounds great, of course. Wouldn’t you want to know if you were breathing in the virus? Wouldn’t you feel more comfortable sending your children to school knowing they are monitored by the eager sniffer of a BioCloud? Don’t you want to put anxious people at ease so we can all return to work and “get the economy going”? BioCloud and its possible use cases are very compelling, which is precisely why the device should be examined with a skeptical lens.

If there is one lesson I hope the public has learned from this pandemic, it’s this: public health is not simple. From issues of basic science to questions of how interventions play out in practice, our assumptions and intuitions are contingent on science, which is often complex and even clumsy when it comes to large scale public health interventions. Many factors that might influence outcomes are not trivial to investigate, and even coming up with a model that takes such factors into account is a daunting task, typically refined by iterations of research across the scientific establishment.

With the real life complexities taken into consideration, I will attempt to address how we might answer the question: is the BioCloud product a viable public health intervention? We can decompose this question further into two primary concerns:

  1. Does the BioCloud device function as advertised?
  2. Assuming a working product, is it an effective solution to mitigate the spread of the virus?

Deciphering the hype

At time of writing, there are no publicly available demonstrations of BioCloud units, no peer reviewed publications, no regulatory approvals, no public health endorsements, and no approved patents. The device is not authorized by Health Canada, which means it cannot be used as a medical device to diagnose. With no information available from independent institutions or regulators, info about the device must be derived from publications by the company and the largely uncritical media frenzy that has followed.

Focusing on the detection capabilities, let’s begin at a high level with relevant technical specifications of the device as described throughout media releases and marketing material:

  • The device is to be installed in indoor spaces up to 1,000 square feet or 225 cubic meters (the white paper also provides detection time calculations for rooms up to 2,500 square feet).
  • The device samples the air in “real-time” (5 times an hour) to detect the SARS-CoV-2 virus.
  • The “advanced” sampling technology “optimizes” the air for analysis (whatever that means).
  • The device uses a proprietary detection chamber. Presumably, the undisclosed patent applications relate to this tech. There are apparently 3 USA and 1 Canadian patent applications that have been submitted.
  • The device uses “three independent [virus] capture techniques that allows for intact virus sampling while achieving a high capture ratio.”
  • The device uses “both a viral collider and a chemical process to trap virus particles.”
  • The chemical process involves an unnamed reagent that is currently sourced from the US.
  • The SARS-CoV-2 virus is ultimately identified with a laser sensor.
  • Lower detection is based on detection of the live virus at 0.005ng, though laboratory tests allegedly found a sensitivity to 50 virus particles at the lower end.
  • As far as I can tell, no human interaction/maintenance is intended to be required during operation with the exception of cleaning and detection chamber replacements, which occur 3 times a year, or following a positive detection.
  • Kontrol provides estimates of detection time that range from 6 minutes for a 1,000 square foot space to 15 minutes for a 2,5000 square foot space (based on a number of assumptions).

I have excluded specifications relating to connectivity/alerting as these are technically trivial and uninteresting (though someone may want to follow up with the possible security concerns of a such a high stakes device supporting numerous network protocols).

According to the BioCloud reference white paper, independent lab validation was performed by “leading virology and microbiology experts” at the following labs:

  • Heinrichs Lab (Dr. David Heinrichs)
  • Dikeakos Lab (Dr. Jimmy Dikeakos)
  • ImPaKt Lab

Dr. Heinrichs himself is quoted as saying “There’s no doubt in my mind that this technology can quickly and effectively detect an array of airborne pathogens, including the virus that causes COVID-19. Our results are absolutely conclusive.” It is not entirely clear what Heinrichs is referring to when he says “this technology.” I reached out to both David Heinrichs and Jimmy Dikeakos, but received no response.

While reports indicate that testing has been performed on a fully operating prototype, this does not appear to be the case based on how the results are reported. This may be due to inadequate documentation; overall, the experimental descriptions and supporting images are unclear and insufficiently detailed and labelled. Personally, if I was demonstrating a scientific achievement with an intent to commercialize, I would make sure that I communicated the important details in such a way so as to render the results unquestionable.

If you scroll to page 15 of the white paper, you will find the opening page of the publicly available findings related to testing of the device. The following page documents the detection process: a “viral target” captures SARS-CoV-2, a reagent is introduced that binds to the virus, then a laser is used to detect the presence of the virus. The process makes sense and is well illustrated within the accompanying figure, but the subsequent experimental documentation raises concerns.

In short, the experiments seem to indicate that a nitrocellulose membrane coated with a receptor protein is repeatedly washed with samples from the air as well as the reagent. After the “the reagent detection sequence,” a “measurement sequence” is performed (no details given). Presumably this last step would be the laser’s role, but throughout the experiment, the paper was only ever said to be “inspected”. There was no laser described or shown.

Considering the absent laser component, a fully working prototype did not appear to have been tested. In addition to the laser, the device would presumably need to recycle the requisite testing materials for sustained use. This would include the nitrocellulose, receptor protein, reagent, and the various solutions utilized. To operate as advertised, this process would have to occur over and over again within the machine without human interaction.

Based on the replacement frequency of the detection chamber, and assuming that the machine operates 5 days a week for 8 hours, the BioCloud would have to perform over 3,000 tests without intervention. In my opinion, this is the more remarkable feat than demonstrating a one-time detection event, but there are further challenges still when attempting to align the testing with the claims made for the device.

Even if we assumed that the device was fully functioning and could perform sustained sampling and testing of the air, we have no idea how it will perform in the real world. Whereas the lab presumably provided a controlled environment with limited air particulate, the intended deployment environments are surely not quite as ideal. How does the device remain reliable for extended periods of time when cycling through the air? Are there conditions that may result in false positive detection events? How does the device ensure the extended lifespan and integrity of the compounds/materials it uses? These questions are unanswered.

Even the calculations estimating the detection time (extrapolated from ideal testing conditions) require a number of assumptions unlikely to hold true in deployment scenarios. In particular, the calculations rely on the presence of a model of a COVID-19 “high emitter” who breathes regularly and coughs once per air exchange. We will return to this assumption later, but it is worth noting that this excludes low or even moderate emitters.

The concerns raised up to this point, if left unaddressed, should be enough to preclude the deployment of this device as a public health intervention. Not only can we not be sure the device fully works, but there is no firm evidence that it can function effectively within environments it is intended to safeguard. Nevertheless, let’s go a layer deeper as devil’s advocate and speculate on the value a fully validated (functioning) device of this nature might offer.

(As an aside, it is also worth noting that there are some mistakes in the white paper that would have been caught by a seasoned reviewer. As perhaps the most egregious example, one of their references [9] lists only a title and a school with no indication of the format or how it was retrieved. It does not appear to be available online and does not appear to be an actual scientific publication, which is unfortunate, as it is referenced in support of a strong claim.)

Life in the real world

Public health considerations are often more complex than they might seem at first. In part, this why public health positions on mask use evolved over time. In addition to questions about how the virus spread and how effective masks were when utilized by the general population, there were also initial concerns about shortages for medical personnel, which could have had drastic consequences. There are further concerns about the use of improvised or inappropriately fitting masks, which may give people a sense of security without offering significant protection to themselves or others. Even knowing that respiratory droplets are a potent transmission route for the SARS-CoV-2 virus and that face masks generally mitigate the travel of such droplets, it was difficult to estimate the benefit at a population level because of the complex factors surrounding the use of masks in our everyday lives compounded with uncertainty of viral transmission specifics. The possible benefits the BioCloud might provide are even more of a mystery at this time.

First, we have to contend with the fact that the BioCloud is perhaps least likely to find itself in the idealized situation such as those at the foundation of the white paper’s calculations. Recall that one of these assumptions was the presence of an actively coughing COVID-19 “high emitter.” So-called high emitters would undoubtedly stand out in a COVID-aware world. Even without a cough (which alone would defeat the assumptions behind the model), workplaces, businesses, and many other places where people gather increasingly screen for symptomatic individuals and may have a strict policy for people experiencing even one of a number of symptoms.

In areas where such screening occurs, we might expect to see spread primarily from asymptomatic people or people whose symptoms are not visible. While such individuals can spread the virus, they are likely not quite as efficient as symptomatic carriers. Of course, this presents an additional impediment to the possible performance in the real world.

One term oddly absent from the analysis and assumptions regarding this technology: masks. Masks have become commonplace in indoor settings where people gather to work in close spaces. While this isn’t universally the case, I suspect it represents the majority of locations considered. Since we know various types of masks mitigate the amount of expelled virus, it is safe to assume that mask wearing would further impede the likelihood of a unit from detecting the virus.

With these few realities taken into consideration, the likelihood of the device actually detecting the virus in a short time in practice is looking to be questionable, but there may nonetheless be some situations where it gets just what it needs: a dose of unmitigated virus right into its detection chamber. Such a situation would include indoor spots where people gather without full compliance to public health guidelines and without direct supervision or screening.

Two possible hotspots that came to mind were certain churches and mass transit such as the Toronto subway, but the former is likely to not buy in to detection technology (even if it worked) and the latter is of questionable value when people are either stuck on a train or moving around the terminals. With people constantly on the move, what value is a detection event that lags 6 minutes at best? What are operators to do when the sick individual has moved on and the virus has already been spread?

Let’s once again – for the sake of argument – dive to a deeper layer of hypotheticals and assume that there exist indoor spaces that satisfy all the assumptions behind the product where both uninfected and “high emitter” individuals congregate for long periods of time. What happens when the BioCloud unit detects the virus?

The team behind the device has indicated that the decision making process following a positive result is up to the management at that facility. So what, then, is a facility to do upon detection? Do they evacuate? To what extent? Is everyone in the facility tested? What happens over the subsequent days? What liability might a company take on with respect to their response?

These are just some of the questions that should be addressed not just prior to deploying the device, but in preparation for conducting an actual clinical trial to demonstrate value. After all, the BioCloud is backed by claims that it can help stop the spread of the virus. To prove that, we would need to study a fully working unit in a real setting with well-thought-out procedures. But again, for the sake of argument, let’s assume that all this work had been done and a study was published showing at least modest reduction in positive cases where units were deployed. We would then ask: is it worth the cost?

All health interventions must consider both the benefits and the costs. We could, for example, perform a COVID assessment (nasal swab) of every Canadian, 3-times per day. We would find more cases, allowing us to quickly isolate those infected and mitigate spread. Would it be worth the cost? I’m no health economist, but I am going to suggest that it would not be.

The BioCloud is priced at $15,000 (on the low end) with a maintenance cost likely around $2,000 per year. Is it worth the cost? The answer depends on a great number of factors that we simply do not know. This makes it a rather high stakes bet, especially for potentially high risk settings that are already strapped for cash.

Where are the scientists?

For a novel technology like this that appears to be the first-if-its-kind on the market, there must be scientists behind it, no? Oddly enough, I could not identify an employee of CEMSI with what I would consider a sufficiently relevant scientific title or background. Just take a look at the LinkedIn listing of employees (disclaimer: LinkedIn is never a comprehensive employee list).

While the company does appear to have some experience with technology that makes up components of the BioCloud, the key components that drive the detection process cannot seemingly be traced back to a laboratory or scientific group. This may seem like a strange way to assess the validity of the technology, but tech of this nature often possesses a scientific lineage that consists of basic science research, early prototypes, publications, etc.

It may be that the company has found a very creative way to repurpose existing technologies into a fully automated product. It may even be that the company has developed entirely novel processes that have evaded scientists working in this area. I am open to being proven wrong, but I am skeptical of these possibilities. Don’t get me wrong, I love the idea of people outside the scientific establishment inventing novel and practical science-based tools, but history is not necessarily in their favor.

I do understand that there exists motivation for secrecy. If the company does have a brilliant product, it would not be in their interest to reveal sufficient information to permit reverse engineering, especially as they do not yet hold patents and there is no guarantee that their applications will be granted. That said, I would be much more reassured to see public support for the product beyond that coming from the company and its apparent shareholders. If you search Twitter for BioCloud or $KNR, for example, you will see unrelenting promotion from accounts that appear dedicated to boost the company’s stock. Of course, I am not claiming that these are bots or paid accounts; they might just as well be investors with a stake in the company. The accounts range from clearly dedicated investors to accounts that have been inactive for years before taking a strong stance on the product and company.

As mentioned earlier, David Heinrichs is the only researcher who was quoted in support of the technology, yet I cannot find any statements he has made about the device besides the primary quote repeatedly attributed to him. I find it odd that he wouldn’t simply respond to my email reassuring me that the device works and that he stands behind his claims and testing. He may certainly be bound by some form of non-disclosure agreement, but I still find it odd that he wouldn’t confirm this restriction or even affirm his published comments.

When I reached out to other experts in this field, they were skeptical, noting the device’s white paper did not contain sufficient information to determine whether the BioCloud worked and how. Others on Twitter (an investor and financial crime expert) shared similar concerns about the lack of public scientist support shortly after the device was first announced. The most relevant credentialed public support for the device I could find was a Newfoundland-based former dentist whose apparent sole tweet (at time of writing and assuming that this is not an impersonator) appears to be no different than the hoards of presumed shareholders spamming social media in a likely attempt to bolster their investments.

Just as curious as the development, one manufacturer contracted to actually make the devices (or at least components of them) is OES Manufacturing – another company in London, Ontario that specializes in making sports scoreboards and wire harness quality assurance devices. They appear to work with a variety of industries, presumably in the design and manufacturing of circuit boards, but it is not clear that they possess sufficient capabilities to manufacture what might be required for a functioning BioCloud device. Then again, without sufficient technical information, it is impossible to know just what the manufacturing requirements are and we do know that some components, such as the reagent, are externally sourced.

I reached out to Brad Young – the BioCloud Technical Manager. My first interaction with Brad was on Twitter, where he wouldn’t confirm with me that there existed a fully functioning prototype, instead posting a picture of some units that did not appear finished. In his defense, he is not very active on Twitter, but I find this a strange question to leave unanswered, especially after initially engaging me. I followed-up with an email to Brad, asking if he would be willing to answer some questions or put me in contact with someone who could. I never heard back.

Skepticism, but not absolute

It should be clear from my writing that I am highly skeptical of this product. I am skeptical that it works as intended. I am skeptical that it can perform sufficiently in real world scenarios. I am skeptical that it has value as a public health intervention and I am skeptical that its cost would be justified in many of the target settings.

I am equally skeptical – as everyone should be – of companies that hype novel technologies without providing sufficient supporting evidence or demonstrations. When it comes to publicly-traded companies, shareholders don’t just stand to benefit from revenue, but from speculation on the value of the stock. I don’t possess the expertise to analyze the business elements of this situation, but others have suggested some interesting trends. Then again, a company can hardly be blamed for capitalizing on opportunity.

Nevertheless, I am not so skeptical to conclude this product definitely cannot work. I won’t even claim that it has no possible value as a public health tool. As a science advocate, I simply demand a higher level of evidence. As a taxpayer, I can only hope that the Ontario government was provided with such evidence.

One thought on “There’s Coronavirus in the air. Can it be detected in real time?

  1. ClashofScience Reply

    I read their brochure and it’s like a promise to the most basic kind of laboratory immunoassay (ELSIA), and has no details how they’ve managed to make it automatic whatsoever.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.