Ronit: Hey everybody. Welcome to BreezoMeter’s live webinar - 7 air quality data trends businesses need to know. Before we get started I just wanted to go over a couple of housekeeping items.
Today's session will be recorded and emailed to you afterward. There will be a Q&A session at the end of the webinar. To ask questions please use the Q&A panel that can be found when you click on the QA button at the bottom of your zoom screen. We will have time at the end of the webinar for your questions, and if we don't get if we don't have enough time to get to the questions you will follow up a few afterward.
On the webinar’s agenda today:
- Addressing air pollution and its health effects
- Why your business is paying attention to the air
- The challenges with air quality data integration.
- Major trends your business should know
- Use cases
- The benefits of air quality data
- Your questions
Our hosts today are Dr. Gabriela Adler, the chief scientist here at BreezoMeter, and Uri Schechterman, head of business development and strategic alliances of the Americas. Take it away, Uri.
Uri: Thank you so much, Ronit and the marketing team, for arranging this great webinar, much appreciated. Let’s cut straight to the chase. A little bit about us: BreezoMeter is a software company that uses big data and machine learning to provide you with hyperlocal air quality and pollen data, and we offer that data to numerous industries. Now I think most people on this call are probably aware of the fact that air quality is invisible, which makes it a major challenge to us. But what we try to do is really democratize the environment and offer people visibility and also simple intuitive action items that they can take in order to do something about mitigating the damage from air pollution. Because let's just use a very basic example:
Pretend right now that the outdoor air quality is bad. If I know that, we can just close the windows and doors turn on the air conditioning, and that in and of itself would actually mitigate 95% percent of the damage from air pollution. So there's a lot of little intuitive steps that we can make to make a difference in our lives, and we provide that data through our API in the various features you see here on the slide. Whether it's pollutant concentrations or pollen data, so people can take those simple actions and do something about it.
But before we get started about the different trends of air pollution maybe we'll just go to the basics and understand exactly what is air pollution. Gabi, maybe you can just give us a quick overview of that.
Dr. Gabriela Adler: Yeah, so air pollution is a mixture of both particles and gases that at high concentration could be harmful to us. Air pollution can come from either anthropogenic sources, meaning man-made, such as factory emissions or traffic, and it can come from natural sources, for example, volcanic ash or mineral dust, or pollen.
Uri: Perfect so let me just go make sure a quick illustration. As I mentioned before the people are aware of the fact that air quality is invisible, but I think what people are not so aware of is just how hyperlocal air quality is and how dynamic it is.
You're about to see a quick video of London. This is the heat map of the air quality. Red is bad. Green is good. You can see the time ticker on the top left corner of the map over here, and you can see how quickly it changes throughout the city. The various areas within the city show completely different air quality data. Streets as far away as one or two miles away from one area could be really bad conditions and another one could be really great, and in the morning could be terrible in two or three hours later it could be amazing air quality conditions.
I think this image or this video kind of tells a lot of words because government monitoring stations are pretty much before BreezoMeter came about, was the only source of reliable air quality data. And those government monitoring stations are really great. They cost a lot of money and it costs even more to maintain, but they give an accurate picture. But just two or three miles of ways from those government monitoring stations can be completely different air quality levels and that's where we come into play. So, a lot of it is really based on the local air quality. For example, traffic affects air quality, local weather conditions affect air quality, factories, etc, all those local conditions affect the air quality, which makes it so dynamic and hyperlocal.
But beyond that, there’s actually a bit of a global challenge as well. It's not just about the local air quality conditions, it's also about the global conditions. Maybe Gabby, being a “rocket scientist”, certified with Ph.D. in atmospheric science, can give us a bit more of an inside look as far as the global challenge that we're facing with air quality.
Dr. Gabriela Adler: So I think most of our listeners would be surprised to know that air quality is not only affected by the local sources, which is quite intuitive, traffic or local emissions but also, it can come from far far away. This video, done by NASA, shows you a type of pollutant called particulate matter and it contains different types of particles.
So let's focus, for example, on the orange in the map which is mineral dust. It's not dust that you have in your apartment, but rather dust that comes from a desert. In this case, you can see how the Sahara Desert transfers all the way, crosses the Atlantic, to Brazil, which is quite a long distance. What that means is that if you are located in France and you see the dust on your car's windshield it can come from the Sahara Desert, which is quite far.
Uri: That's amazing, and what I love about this particular video, if you look at it's almost like you see the equator quite clearly. I mean just north of it you see the sandstorms just described, to the south of it we do see all these little green and white circles over there. What are those exactly?
Dr. Gabriela Adler: So if you look at the legend it would say ‘organic carbon’ and ‘black carbon’. Let me put it in words that everyone can understand: that means fire, sort of massive wildfires, in Africa and South America.
Uri: That's just amazing because again, people in America can be suffering from sandstorms that are happening in Central Africa, or from fires from there. Let's talk a bit more about air pollution, and before we go further about the trends, I think people are familiar with the World Health Organization's numbers about that. It affects virtually the entire world's population. The cost of it has seven million deaths, which is more than HIV, malaria, and car accident injuries and deaths combined. So at the end of the day, we know the macro-level, but maybe we should just focus a bit on the individual. What does it do to us? How does air pollution really affect us, Gabby?
Dr. Gabriela Adler: So the first question we need to ask ourselves is: who is ‘us’, right? Who is this general public? Are you a senior? Are you a pregnant woman?
Uri: She's pointing to me when she says ‘pregnant women’ but that's just a joke, okay?
Dr. Gabriela Adler: Are you a child? An athlete? Air pollution could affect you differently depending on that. The effect of air pollution could be divided into short-term effects and long-term effects. Short-term effects, are things that you experience on a day-to-day basis, which you might not be aware of, like headaches and coughing. The severity is highly dependent on who you are. One person could cough while the other could have pneumonia.
Long-term effects are a bit scarier, I would say, and less talked about. They include Alzheimer's, heart diseases, fertility problems for both men and women, premature labor, and I even saw a study discussing the connection between air pollution and happiness.
Uri: Okay, so I want to keep this actually on a happy note because, again, there's a lot that we can actually do about air quality in terms of mitigating the damage from it. You mentioned the word happiness and I just want to talk to you about specifically one of the first newest cases that we actually had, and that was something that we did with Mount Sinai Hospital.
The idea was they created an app for their asthma patients and the idea was to see whether or not, once the asthma patients are aware of outdoor air quality conditions, would reduce asthma attacks and hospitalizations or not? The interesting thing was the answer was a resounding yes, which would kind of anticipate, because we perceived that if somebody is aware of the outdoor air quality conditions, and you can make it visible for them and intuitive with the call to action, they would do something about it. And obviously reduce whatever, you know, mitigate the risk from it.
But what we were surprised to see was really the number of times that actually they spent outdoors. Because when the study started, we were very concerned that people would look at the information about the air quality or air pollution as something which is scary, maybe like a diet app, a restricted app, don't eat this, don't drink that, pace yourself.
But in a sense, it was actually the exact opposite. It was more like a liberating app. Meaning, it enabled them to spend more time outdoors, because, by definition, the people in New York City, I guess those asthma patients presume that the outdoor air quality was bad all the time, where in fact, BreezoMeter, through our API, could actually let them know exactly when the outdoor air quality conditions were actually good and they could spend more time outdoors. They can pick the times of the day they can spend more time outdoors. They can decide they want to go for walk two blocks east as opposed to two blocks west.
So the goal here is really to not restrict people. It's quite the contrary. The goal is really to allow people to live their lives, to liberate them, to make them aware of the air quality conditions, but engage them in such a way that they can live their lives to the fullest in a healthy way.
Maybe what we’ll do, we’ll just kind of jump now straight to the trends that we were talking about earlier, and see various examples of these trends that we've seen over the last year or two. Because, if we look again, people were aware of air quality a couple of years ago, two-three years ago, even further back, but they didn't quite, they weren't necessarily aware of all the ramifications of it.
The first trend we're gonna talk about a bit is the real-time trend. I'm using the weather app example here intentionally for that reason because the weather is something that we're all familiar with. Most of us check it maybe once or twice a day I guess, depending on where you live around the world. But air quality is far more dynamic than that, and as a result that people actually have to check air quality more often than that.
On aggregate, we see from what we call wellness app or weather apps or clients like that a major increase in user engagement and times checked, simply because people are coming in to check and understand what the air quality is right now in real-time. So it's an increase on average of about 2.4 times on aggregate for many of our customers in that regard. And that's really the key because air quality is dynamic. It's not like the weather. It changes much more often.
The same can be applied to another trend which is: location-based. Whereas in the past people were looking to get information on a citywide level for air quality, it's almost irrelevant to do it in such a way. As we saw earlier in the image of the City of London, and all of you guys are welcome to just check out our heat map on our site and see for yourself, different areas within the same city have completely different air quality conditions. So location-based is crucial here. One of the examples that we have is with some of our clients who are in the air purification and HVAC space. They can notify their customers and let them know exactly what the outdoor air quality conditions are, and also showcase their value proposition.
The indoor air quality is great, the outdoor air quality is less than great. You can see the difference in the HVAC or air purifier and how much of a difference it makes within the house. This is just an example from Blueair. We have that with Dyson and Johnson Controls, and many of the top leading brands when it comes to that. But again, more than that, one of the things that we see in terms of the IRI is that people are buying more devices as a result. Ironically, some devices that may have been bought in the past, without being aware of the outdoor air quality data, will literally be collecting dust.
Because if somebody will open their window look outside you see that it's a beautiful brisk day, they may think and decide that, you know, the air quality is great. But again, it's not really available. That information is incorrect, where in fact the air quality can be bad. So as a result, they are turning on the air purifier and using it. Renewal sales went up and filter sales as well.
To me, I think personally, one of the most exciting changes that we're seeing is really the change from people relying on the AQI and moving on to the pollutant concentration. I'm gonna have Gabby elaborate a bit more about this because it's an important topic. People have been on misusing, I guess, the air quality index to a certain extent. The air quality index is typically defined by the worst dominant pollutants. Most countries don't have it as an average. They determine thresholds for each pollutant and based on their interpretation of those thresholds they determine what's the worst dominant pollutants and therefore the air quality index is defined by that.
So you can easily see situations wherein one country certain air quality conditions are defined as good, in another country the same identical air quality conditions would be defined actually as bad or even moderate because it's just cynically manipulated. So maybe Gabby, can you just give us a bit more information about the pollutant concentration element of it?
Dr. Gabriela Adler: Yes so as you've said rightfully, the air quality index could be political. But even if we look at the countries whose air quality index is based on epidemiological studies, which means medical studies connecting air pollution to hospital admission, these studies are not specific enough. What does mean? If I'm a pregnant woman I don't care only about the dominant pollutant. I care about the second dominant pollutant and the combination of the different pollutant concentrations.
Uri: Perfect. So it's a perfect segue to the next trend that I want to talk about which is more personalized or, I call it the intuitive or the personalized, it’s the call to action, but call to action based on segments. As Gaby just mentioned. air pollution has different effects on different sections of the population, and I'm using ‘section of the population’ as an example because ‘demographics’ is not the right term for it.
We have identified 7 types of population segments that could be affected by air pollution. Let’s divide them into the general population, the elderly, the young, people with heart disease, people with lung disease, pregnant women, and athletes. I'll talk more but athletes later, but athletes actually are one of the most unappreciated in terms of understanding how bad air quality affects them. An athlete running outdoors is inhaling much more air than the average person. So they too need to be on alert to understand what the air quality conditions are.
We can personalize the recommendations as you can see here. Different segments of the population get different recommendations based on the specific pollutant concentrations, not just on the air quality index. But in addition to that, it could be good recommendations like, go outside, time to go outdoors, the air quality is great. And by the same token if the air quality is not as good, warn them, letting them know what they can do to mitigate the damage being expected by the outdoor air quality.
Another very nice example that we have in, I guess, personalization in action, is the connected inhaler that we have a partnership with KapCode. KapCode basically uses BreezoMeter’s air quality data to engage its users. What they're doing is letting their customers know exactly what the outdoor air quality is. Based on that, these asthma patients can then make a determination of what they need to do. It could be very intuitive. Tell them to take certain preventative measures of medicine or inhaler, whatever it is.
But better yet: this is on the engagement side. On the analytics and research and development side of the business, they're also able to understand what happened prior to every time the client or the customers use the puffer. So if somebody uses the inhaler they know what preceded it. They can see the correlation between different pollutants and different concentration levels, so there's a lot of R&D element here in terms of correlation and cause and effect that a lot of the digital health analytics companies as well as pharmaceutical companies are finding very helpful over here.
Another trend we see is data visualization. A heat map tells a picture, and the colors of red and green tell a picture here. Often that is the segue that kind of introduces people to air quality, to make them aware of what it is. Another very nice example we see here is actually with AllTrails. AllTrails is the leading app and website for hikers and those who want to do trekking and biking outdoors, and they have a heat map overlay showing exactly what the outdoor air quality conditions are on a map in a visual way.
So people can decide what hikes they wish to take. Again, due to the hyperlocal nature of it all, and due to the activities that are endured by athletes that are inhaling more air than the average person, they are particularly sensitive to outdoor air quality data, and therefore it's important for them to know and better understand which track they can be using.
Let's move on to the IoT automation side of things. So we spoke before about understanding the effects and triggers of air pollution on a personal level. But whereas in the past those recommendations were done something along the lines of a trigger or recommendation of sorts now it can be automated. Here's a couple of examples that we have of that:
One is with Delos that creates a smart home environment. Their whole business is based on smart homes, creating wellness-related homes. They can decide, based on certain triggers, to activate things in automatic ways. You can do notifications and alerts and turn on the air purification system as a result of it could also be automated based on specific triggers.
So if certain pollutant concentrations are high, the recommendations can be to turn on the HVAC system in the house, do it on indoor recycling as we see here on the glass automation, the glass device from Johnson Controls. But you can also do it manually. You can decide on your own what triggers it and obviously control a lot of those things from an app not just from the dashboard inside your house.
Let's move on to maybe the last trend. Before I'll give the example here, Gabby maybe you can give us, you know, we’re familiar with the numbers overall about pollen, which I guess is more of a natural pollutant, but it affects 50 million Americans, a lot of money spent every single year on allergy medications, and it's increasing the chance of asthma. Can you give us a bit of a connection between air pollution and pollen?
Dr. Gabriela Adler: Yes. So the first thing is that pollen is a particulate matter, so it's a type of air pollution. That's the first thing. And the second is the allergic reaction. Pollen can be enhanced in the presence of air pollution.
Uri: Perfect. So let's give you one very simple example. This is a nice example that we have with ALK, which is a pharmaceutical they're running right now in Europe. It's an app, it's an app meant to alert people who suffer from pollen allergies to exactly what the outdoor air quality conditions are in terms of pollen. So you can see what are the dominant pollens right now?
Some people are only allergic to birch, others are allergic to olive. Different people have different types of allergies so we do break it down to ten (now 13) types of trees and grasses as well so based on the various sensitivities they can control the pollen allergy and see what they can do in order to address the particular effects, irritation or allergies that they were getting as a result of the pollen.
Now, one of the most difficult tasks we have here, we're getting a lot of questions from the audience about this, is explaining our technology, something that we worked on for the last five years. A tremendous amount of big data and machine learning has been put into this, so Gabby, you got the pleasure of trying to sum up the last five years of 40 different development engineers into two minutes, good luck with that. I'll try to see what I can do to help. It's not easy to do, so go ahead.
Dr. Gabriela Adler: Okay, so let me try and walk you through it. Let's start with the monitoring stations, the government monitoring stations that Uri mentioned before. They're expensive equipment that is well-maintained measures different types of pollutants. So they have different sensors, and we collect data from over 41,000 sensors worldwide, more than 80 countries. It’s pretty intuitive how we use this information. Now we also use weather, which can be, I'll give you two examples, one is intuitive the other is less intuitive.
So the first point is rain. We all feel that air quality is really good after rain and it's true actually. Rain cleans out particulate matter. It cleans out pollen as well. The non-intuitive example is the sun. Ozone is created in the presence of the sun. So I can be outside on a really nice sunny day, but it's really polluted outside, and I cannot feel it because I cannot smell it and I cannot see it. So that's another example.
We also use dispersion models. These are models that predict how the pollutants move in time and space from one place to another. We also use traffic data. We collect the data every 13 minutes and we connect between the traffic and air pollution using machine learning and it's I think quite intuitive to understand, that traffic actually affects air pollution. The other thing that we do is pollen, which is a bit more tricky since there is a biological factor to it. We need to put into our models and phonological factors, seasonality, emission of different types of vegetation, the land cover, etc.
So all this information goes into our algorithm and we first do quality assurance to make sure that the data is valid. Then we use statistical analysis and machine learning-based algorithms to provide you with two APIs, one for air quality and the other for pollen.
Uri: Perfect. So again those APIs are basically then integrated in any way the customer wants. For example, some people are talking about the business applications here. It could be used on mobile devices and watches and apps, also on the website. As I mentioned, a lot of these pharmaceutical companies and health analytics companies are using this data also to do a lot of correlation work behind the scenes in that regard.
The one thing I love about this particular business that we're in is that people always find new ideas and new uses for our API in ways that we never imagined. Most recently we signed up L'Oréal, who’s now using our API for skin care applications. That's something if you would’ve told me two or three years ago that L'Oreal will be using our data for skincare products I probably would not have imagined that. But that's something that's done. I know there's a lot of innovation people sitting there listening in to this conversation, we're just sharing the technology and what it can do but it's really oftentimes the best creative ideas come from the product managers and innovation managers that come up with new ideas and new ways to integrate our data.
Maybe just to wrap things up before we go to the takeaways and the Q&A: What are the benefits? There's the obvious benefit of engaging the user, of making them aware of outdoor air quality data, but there’s a lot we can do to mitigate the damage. If you're talking about an IoT device, whether it's air purification or HVAC, again just turning it on when it's polluted would mitigate 90% of the damage.
We've seen major increases in sales for those IoT-connected devices, both in terms of unit sales but also in terms of filtration sales and service fees. In terms of just user engagement, on an aggregate level, if you look in apps that have existed before but just introduced the air quality features, that increased engagement on average by 2.4 times. Specifically, if you're looking at the heatmap alone, we see a greater and greater increase in heatmap usage, and now the average session for the heatmap usage is now 44 seconds. So before I wrap it up and give it back to the Q&A to Ronit. I just want to say thank you again to Ronit and to the marketing team for doing a great job putting this together. Much appreciated. Gabby, I think you have a parting thought for us as well.
Dr. Gabriela Adler: Right. I wanted to add that the average adult inhales seven to eight liters of air every minute, and that sums up to 11 thousand liters of air per day. So Ronit and Uri, thank you and let’s breathe well.
Uri: Amen to that.
Ronit: Thanks, Gabby. Indeed, let’s breathe well. Thanks, guys. Just want to have one more look at our takeaways from what we learned today.
Businesses have an opportunity to make a positive impact because as we learned, air pollution affects everyone. Air quality data monitoring relies on proper technology and the proper integration of air quality and pollen data can increase user engagement and ROI. Now as we come to a close, we want to take a couple of questions if you guys are up for it. Fabulous. The first one we have is:
What exactly goes into integrating BreezoMeter’s API?
Uri: Okay so I guess I'll take that one. At the end of the day, a lot of it depends on how the client uses the data. The integration itself is very quick and I think the average once they try the API they integrated the onboarding is usually done on a pilot phase, we're talking about one to four hours ma. That's how long the integration itself takes. Oftentimes if they come back with some specific questions it could be a bit longer, but it's a matter of hours to a day or so. And the API codes are very simple so for example, if somebody has started with us in a particular country but they want to expand to other countries it's the same API code, same developer keys, all they have to do is, there's a field that we open up to add additional countries. So obviously the integration is super fast and super simple.
Ronit: Great, thanks Uri. One more question before we go: how do you guys check your accuracy?
Dr. Gabriela Adler: Okay so I'll take that one. Uri if you have something to add please do. We have a continuous accuracy testing system and it's really important for us to test our accuracy continuously. Since we cannot deploy measurements to test our accuracy all the time and worldwide in more than 80 countries, what we do is use the ‘leave-one-out method.
What is that? Imagine that outside of your office there are five monitoring stations and you use all of them to put in your algorithm. Now I take one off and I try to predict what is the air quality in the location of the station without using the data of the station. And then I compare the actual measurement from the station to what the algorithm predicted.
Uri: Perfect. Okay, I couldn’t have said it better myself. That's amazing. Another question that we saw here is about, you want to maybe read me the question? I can’t see it as well right now.
Ronit: Have you applied machine learning for AI to forecast air pollution?
Uri: The answer to the questions is yes. Machine learning is what we use to do that. I was explaining to my kids the other day, how the machine learning process works.
So basically, if you look at the raw data, one of the first things we do is actually validate the data. We don't think the government monitoring stations sensors straight into the algorithm. We first want to make sure that the information is correct, and oftentimes up to let's say 12% of the time it could be incorrect because of, you know, it could be that it's frozen feed, it could be that the sensor hasn't been calibrated, and after that, the data then goes into the whole machine learning processing and algorithms that are used over there.
So it's almost like the way I explained to my kids: you want to make sure, if you're cooking a good meal, you want to make sure you have great ingredients. You can be the greatest cook in the world but if the ingredients are not good it's not going to be very good. So we took care of the veracity of the data before it actually goes into the algorithm. That's when we run the machine learning validation. And after we run the machine learning validation on the data or ‘the vegetables’, whichever way you want, if I took my cook example, then we run the algorithms on it and build it up from there.
Ronit: Great. Thanks, Uri. I just wanna take one more moment to thank our speakers today, Gabby and Uri. It was great and informative. And I want to thank all of you for joining us today for our webinar live event. Can't wait to see you all at the next one.