Scanner monitors vitals in seconds, intoxicants too

TORONTO – Predictmedix AI, a Toronto-based startup company, has produced a Safe Entry Station, an AI-powered scanner that can instantly monitor a variety of vital signs, such as heart rates, blood pressure, body temperature and respiration. The walk-through scanner can even detect alcohol or cannabis intoxication.

As reported by the Toronto Star, the healthcare tech company insists its AI scanner is revolutionary. “We are about to reshape healthcare practices on the global front,” says Rahul Kushwah, co-founder and chief operating officer at Predictmedix AI.

However, as Kushwah explained, the company is more keen on attracting Canadian bosses, rather than hospitals, for now. He spoke to the Star in early September:

What can your AI do that a well-trained doctor or nurse can’t?
If you go into any hospital in Canada, it’s going to take anywhere between 45 minutes to an hour to undergo the whole triage process. We are able to shrink that to less than 20 seconds with our technology, which is able to capture all of those vitals in a completely non-invasive fashion – in a contactless fashion. The results seamlessly integrate into the hospital medical system.

In addition, we have perhaps the only technology out there that can identify cannabis impairment, along with alcohol impairment, without the need for biological fluids. We also just recently advanced our technology to identify if someone is diabetic without the need for them to give fluids. They just have to stand in front of one of our units.

How is your AI able to do all of this without having physical contact with a patient?
When you look at one of our stations – we call it a Safe Entry Station – it has multi-spectral cameras that look across several different wavelengths, with a host of sensors and an edge computing unit: an extremely powerful computer sitting right there on the premises. Our proprietary intellectual property allows those multi-spectral cameras to analyze your face, look at subtle changes in blood flow patterns, and convert that into data on your blood pressure, heart rate, presence of alcohol or cannabis, and so on. Right now, we cover about 18 different parameters.

How accurate is it?
We’re sitting at an accuracy of over 90 percent for all of our parameters.

Just to give you some more perspective on the cannabis impairment side – right now, there are companies trying to develop breathalyzers to identify THC levels. But just because someone has THC on their breath doesn’t mean the person is impaired, because THC has to cross the blood/brain barrier. That’s why, across the U.S. and other jurisdictions, regulators have said breathalyzers cannot be used to measure cannabis impairment.

What we’re identifying are the changes in physiology that happen after THC crosses the blood/brain barrier, and then correlating it back to a person’s THC levels.

You’ve deployed your AI in hospitals in both India and Indonesia. Can you tell me how those trials have gone? Were they the first to see the deployment of your technology?
We’ve also deployed at some major sporting events in the U.S.

But as the technology evolved, we realized that healthcare is truly where there’s an absolute necessity for us, particularly in countries like India and Indonesia. When you look at a hospital in India, it’s not one where the daily flow of new patients is a few hundred people. You have thousands of people coming in daily. You need hundreds of doctors and nurses to measure the vitals of everyone coming in for triage.

That’s where we started to focus in that specific jurisdiction. We had government officials that came here to Canada to look at our technology, and that led to deployments in major hospitals in India. As we speak, there are probably thousands of people that are being screened using our technology, and that data has been seamlessly integrated into hospital management systems.

Have you been in any talks to deploy your AI scanners in Canada, given our shortage of nurses and doctors right now?
We’ve started off with some initial discussions, but we are a relatively small company. We’re focusing on major markets that can offer us that tremendous return on investment. When you look at Indonesia and India, we’re talking about 40 percent of the world’s population, and India has one of the world’s biggest economies. That’s why, for us, India and Indonesia are a tremendous opportunity.

What about the United States? It’s also a big market.
The United States is definitely a big target. But just to give you some perspective – one of the milestones we have recently reached is scanning 200,000 patients. Normally, AI healthcare companies develop algorithms and technologies based on a pool of 2,000 to 5,000 patients.

If we were to go out and do clinical studies to scan 200,000 individuals in North America, we would probably have to go through $50 million to $200 million. India and Indonesia offered us that opportunity. That’s why we’re going full steam ahead in their hospitals. That’s where our team is completely focused right now.

Are you using the data from India and Indonesia to help train your AI?
Of course. You’ve got to remember that the data is being collected in a hospital setting, and it has to be sent off to train the algorithm.

What are the privacy implications of that? Do patients know that their personal data is being used to train an AI?
First and foremost, all of the data is fully anonymized before it is used to train the algorithms. One of the questions that comes up all the time is what happens if someone hacks into our cloud and we lose all of our confidential data. Initially we were using cloud computing, but we switched to edge computing, where all of the computing is happening on the premises where the technology is deployed.

The only communication that would happen every two to four weeks, depending on where the unit is, is to upload patches of new versions of the algorithms. We’re working with clinicians at hospitals where they have consented for us to use the data in those settings. There’s no way you can look at the data and trace it back to the individual.

What is the cost of one of these units, versus a qualified nurse or doctor that’s capable of doing all this triage work themselves?
Our business model is basically a lease model. For $2,000 to $4,000 a month, you get unlimited use of the technology. You can scan as many people as you want.

You mentioned the accuracy of this AI is over 90 percent. A misdiagnosis of a medical issue, or impairment, can have really serious consequences for patients. Who is liable if these machines are inaccurate?
The machines are screening technologies. They are not diagnosing impairment. They’re screening to identify individuals who are more likely to have a substance present in their bodies. When you look at the studies that have been conducted on false-positive rates of alcohol breathalyzers, they can be as high as 50 percent. That’s why the ultimate test to confirm if someone has alcohol in their blood is to give them a drug test.

The technology is more about a top-line screening mechanism. Individuals that are flagged, depending on the jurisdiction, can undergo a secondary drug test.

It sounds like you don’t see this as replacing the human triage process, but rather augmenting it.
When it comes to measuring impairment, yes. But when it comes to measuring vitals, we have been able to get it to accuracy levels where it can really completely replace the entire process of having nurses put a bunch of medical devices on you to measure your vitals.

Are there any Canadian buyers interested in using your technology, from a workplace safety perspective?
We have had very strong leads in the oilpatch. I’m sure you heard about the whole situation with Suncor last year, after a worker died, and the CEO stepped down. In North America, fit-for-work screening has been mandated by the government. If you were to go into a car manufacturing plant in Ontario, you’ll find people there who are keeping an eye out to see whether people are fit to work or not. That is all these companies have right now.

Ours is the only solution of its kind. As much as we’ve had these leads coming in from the oilpatch, we’re a small company with a small team right now. Truly, we cannot capitalize on all the opportunities for us right now. The low-hanging opportunity is the healthcare setting in Indonesia and India, and that is what we are going after.