On this episode of the AI For All Podcast, Richard Barnes, CEO of Select Research and inventor of the Body Volume Index (BVI), joins Ryan Chacon and Neil Sahota on the AI For All Podcast to discuss AI in healthcare and measuring human health. They talk about the applications of BVI, how AI is diagnosing health issues, longevity, measuring health in real-time, using AI with wearables, training AI models with diverse datasets, reunifying medical information with AI, and using AI to understand the health of a population.

About Richard Barnes

Richard Barnes is the inventor of the Body Volume Index (BVI), a digital weight measurement developed over the past 15 years as an alternative to the Body Mass Index. BVI produces over 200 measurements just from two digital images of a person to measure weight distribution, body composition, and much more. AI is being integrated to provide insight into the differences in our body shapes across nations to work out who we are and what we are.

Interested in connecting with Richard? Reach out on LinkedIn!

About Select Research

Select Research has been a pioneer in measuring human body shape for over 25 years. They have managed major national sizing surveys for clothing retailers using 3D technology to measure the outside of our bodies. In 2007, Select patented the Body Volume Index (BVI) to measure the inside of the body for healthcare and this has been developed since then with international academic and scientific partners, including Mayo Clinic.

Key Questions and Topics from This Episode:

(01:10) Introduction to Richard Barnes and Select Research
(01:29) How AI is being used to measure people
(02:16) What are the applications of BVI?
(09:26) How is AI diagnosing health issues?
(11:52) Will AI measure our health in real-time?
(13:57) Using AI with wearables
(15:05) Training AI models with diverse datasets
(16:16) AI in healthcare
(21:20) Will AI reunify medical information?
(23:26) How will AI be used to understand population health?

Transcript

- [Ryan] Hello everyone and welcome to the AI For All Podcast. I'm Ryan Chacon. With me today is my co-host, Neil Sahota, AI Advisor to the UN and one of the founders of AI for Good.

- [Neil] I didn't know this, but as of yesterday, the UN has started calling me the Godfather of AI for Good. I guess godfather is the new moniker.

- [Ryan] I gotta change this in my notes. That's how we're going to introduce you, the Godfather of AI for Good now going forward. I also have with me Nikolai, our producer.

- [Nikolai] Hello. I don't think I can live up to the title of Godfather.

- [Ryan] On today's episode, very exciting conversation we're going to have. We're going to talk about what can AI tell us about our physical health? Can AI be more accurate than physicians? What is it measuring about humans? To discuss this today, we have Richard Barnes, the inventor of the Body Volume Index and CEO of Select Research.

They are a pioneer in measuring the human body, empowering industries like retail, healthcare, and fitness with accurate data. Richard, thanks for being on the podcast.

- [Richard] Yeah, thanks so much for inviting us and yeah, delighted to share some things with you.

- [Ryan] Let's start this off by talking about how AI is being used to measure people. And when we say measure people, being able to look at body composition and things like that and tell things from it. So what does that look like? How does that work? Let's start there.

- [Richard] Yeah, the AI basically provides insights into human body shape that we see with the human eye, but we don't really understand. So what, the way that the body volume index works compared to the body mass index is the body volume index looks at where the weight is, not what the total weight is.

So BMI has some value for populations, but it doesn't tell you what the weight distribution is. So the AI in BVI allows us to work out where the weight is and what that means for your health.

- [Ryan] And with everybody being built differently, how does, how is that data able to be accurately assessed given that fact that people are different. When you're looking at where the weight is, where it's distributed and so forth, is there an, is it pretty consistent regardless of body type, ethnicity, et cetera?

- [Richard] I wish it was. Unfortunately, it isn't. And that's why we need AI to make sense of it because there's sort of some things people know, muscle weighs more than fat. A lot of people know that. So, you can have the same volume of fat and the same volume of muscle. But muscle weighs more than fat, so that's why BMI just doesn't work because people are told you're obese, so you gotta go and exercise more, so they exercise more, they put on muscle, and they lose fat, but the weight goes up. But in answer to question, what happens when you put muscle on for example, and you lose fat, is you lose fat around the belly, but you put muscle on the arms and the chest or the upper legs.

And that means the volume increases in those areas of the body and decreases in the areas of the body where the fat's stored. So, you need to measure that, and you need to understand it. There's other things like we know the weights and the volumes of different body parts, and the one thing about the human body is it's consistent on where the parts are.

The heart's always here, the lungs are always in the chest. The small intestine is in the stomach. So each of those weighs a different weight and has a different volume. So we take all of that into account. But there's only so much a human being or human beings can do. You need AI to make more sense of it.

- [Neil] How many different areas, Richard, are we really looking at when we look at the different aspects of volume?

- [Richard] There's kind of the really obvious stuff. For example, like the chest volume. So for example, under BMI, lots of bodybuilders are told they're overweight and obese when they're not, but their chest volume is going to be mostly muscle. Then around the abdomen, people like me carrying a bit too much around the middle, it's, that's more of a risk because that's called the visceral fat, which is the nasty stuff around the organs.

You've then got some really quite detailed things. For example, like the skin. The bizarre thing about the skin, the skin is about five to six percent of our weight. But it's actually thicker on the upper part of the body and particularly on the shoulders and the upper part of the arms.

Because basically, over thousands of years, the body's learnt to protect itself against the sunlight. Naturally, our bodies have evolved. Even though you can look at the skin being 5 to 6% of the weight, it actually weighs more on the upper part of the body. But obviously having thicker skin on the upper part of the body isn't a health risk.

So, yeah, and then other things like if you've got big hands or big feet. That's fine, but it's not a health risk. So all of those are factors that you need to look into.

- [Ryan] Talk to us a little bit about the application of this kind of technology and where it's being used. How is it being used? And why was this something that you all even ventured down the path to work on and create? Because obviously a lot of listeners, we go to our doctor, they, we walk in for a physical, they weigh us, they give us our numbers and our information, that's it. So how is this being applied? Where is this being applied? Or how should we be thinking about that?

- [Richard] I will answer it, but I think it's worth talking about the alternative, which is the Body Mass Index, and just talking a bit about that first because people understand a bit about BMI, but they probably don't know the full story. And it's worth sharing with you because BMI is just height and weight.

So it's someone's height divided by their weight. So it's basically height is normally a constant in adults and weight obviously is a variable. So you're measuring a variable against a constant. The guy who invented it was a brilliant Belgian mathematician called Adolphe Quetelet. And he invented it in, believe it or not, 1835.

And when he designed it, it was meant to be a population tool only. It wasn't meant to be used for individual risk. So basically it was being used for something completely different to what it was intended to be by the inventor. And I've read his book, and in his book, he actually says, look, this is a start, but it's not the full story, and in the future, science is going to have to evolve, and you have to look at the organs, and you have to look at different ways of measuring the body. But BMI was invented in 1835, and what's so ironic about the timing of this is that I think it was on Monday. The American Medical Association announced a change in policy, and they've said that they believe that the American medical profession should veer away from BMI and use waist circumference and waist to hip ratio and body composition as a better means of evaluating risk. So, it's weird because we, when I started this 15 years ago, I didn't realize how difficult it would be and how ingrained BMI is in the way that people do health analysis. And so that was a really welcome announcement from the AMA to basically say, look, we need to move on from BMI. And yeah, Quetelet, if he was alive today, I'm sure would say, carry on because he didn't design BMI to be anything else than measuring populations.

- [Ryan] Growing up, every time I go in and get a BMI thing, like in college and stuff, it'd always tell me that I was overweight in some capacity. I'd look at myself like, I don't feel like I'm overweight, but obviously we're showing the flaws there, right?

- [Richard] You asked about where it could be applied. It isn't just for doctors, but it's also in healthcare insurance. And the insurance companies use BMI to assess risk. And, without naming any names, we talked to some big insurers and reinsurers in the States and elsewhere. And the really interesting thing is that they've got a problem now where people who are athletic or bodybuilders are actually getting very high premiums because of their BMI being high, and then they're sending pictures of themselves into the insurance companies to say, why have you given me such a high premium?

Because you told me I'm obese, but look at me, I'm not. So as I said at the beginning, BVI is common sense dress up. You can look at someone, and you can look at their weight distribution, and you can think, that person is obese or that person isn't. But we basically use cameras now. We are able to do it from just two images of a person, which is, it's as simple to use as BMI is.

And then the AI kind of makes sense of what that data is and the body composition. And that's the key thing is BMI is so simple to do, you just need height and weight and you've got a number. But now, luckily with the way technology's improved over the years, we're able to do something that, in many ways, is just as simple.

- [Ryan] You mentioned a second ago, you're talking about where the fat potentially is in the body and how that is viewed as either healthy or not healthy, and it's actually we're learning more from that, correct? So how can AI really contribute to diagnosing health issues as well as improve or increase longevity for people?

As we continue to progress and advance this technology, where do you see that going or how is it being used now to really help identify things that maybe weren't able to be identified as easily prior to this being deployed?

- [Richard] The key thing is to say what this isn't. This isn't designed to replace a blood test or replace an MRI scan or actually the evaluation of a doctor who's got 30 years experience of seeing patients with a certain condition. It's designed to give a better start than BMI does, which then will hopefully accelerate the way that a patient is treated better, that it allows a doctor to see signs earlier of where there may be symptoms.

Mayo Clinic are our main collaborators and have been for the last 15 years. And they've just published a paper on 1,200 patients, a 10 year study where they compared BVI against BMI for the prediction of cardiovascular risk, diabetes, and hypertension. And they believe, I think it's about 19% better at the prediction of those diseases, which is a massive improvement, but it still isn't the answer to everything, and the AI will help us see things that perhaps we just don't even know are there. One of the things that's worth sharing with you, because it's just an interesting fact, is that from our analysis of the body and all the measurements that we can get from our tech, we know that in most people, the length of your foot is the same length as your, from your elbow to your wrist if you put your arm at 45 degrees.

Now most people look at it and go, that's not the length of my foot, but in most people it is. And that sort of insight we get from having been experts in this field for a long time, but AI just gives that little bit extra, finding relationships between people of certain body shapes that may be more perceptible to certain conditions.

I wish I knew all the answers, but that's, we can't, you can't analyze that sort of stuff even with human expertise. You need AI to do that.

- [Ryan] Do you see this evolving into something that maybe if not this technology, other AI technology, being able to monitor and measure us as people in real-time and be able to better analyze health conditions more or elements of just things about our health more frequently?

- [Richard] Yeah, I do. The people that we're talking to about screening are looking at screening people every six months. I think I can disclose that, so that will be a track, effectively a tracking study, and that's where it becomes so powerful because if you've got the data on, the measurement data of somebody from ten years ago and then in another ten years time, and obviously today, then you've got the comparison, you're able to then do tracking of the changes in someone's body shape over time.

And, I think it's no secret that, we all shrink. Most of us shrink as we get older. Most of us put on weight when we're older. And so there are changes over and above that. There's other applications like osteoporosis, which is a particular condition of curvature of the spine.

And that is normally done by looking at someone. People of old age who suffer from osteoporosis, which is shrinkage of the vertebrae between the spine, would be able to perhaps predict that a bit earlier because you'd have two scans that you'd see the difference and overlay them and yeah, I think AI could show things that perhaps, either earlier or better or new things that we don't even know are there because the human body is a very strange shape if you think about it. You've got two arms, two legs, a head and this sort of torso and that combination of different things is often very unique to people, and so you need AI to make sense of those nuances that you can't really see.

- [Ryan] I wonder if you could also pair this with data that you're collecting from wearables and other types of devices to be able to have a fuller picture of things and allow AI to be trained to handle different types of data inputs to do exactly what you both are obviously saying here is being able to do early detection, be able to better understand what's going on with the person's body at any given time.

- [Richard] We're using AI to measure the human body better than the human can, but that only provides the data, but then adding that to other data that is AI generated or just available, so how many steps someone's done, how many hours sleep they've had, all the other things that all the wearables and other devices collect.

Yeah, absolutely fascinating because there may well be knock on effects of that on your body shape that we don't really understand.

- [Ryan] Absolutely. And one thing I wanted to ask you because you brought this up earlier about just like the uniqueness and or maybe like the oddness of our bodies. And obviously everyone is quite different. How do you go about training a model to understand the nuances of the human body, the difference between people's body volume distribution, lifestyle, ethnicity, demographics, age, gender, you name it, all these things play into this, right? So how do you all think about that or incorporate that in order for the model to have those data points to understand?

- [Richard] It's taken 15 years, and the majority of that was research and development and planning and getting all the, what I would call the core data right. And then we spent about three or four years creating the algorithms to work out that if the volume changed in a certain part of the body, so for example if the chest increased in volume, how much of an assumption can you make on that that extra volume is muscle, and therefore how much extra weight does that create? And what BVI is really is a weight distribution system. It tells you where the weight is on your body, and therefore what that means for your health.

That's what we ended up with, but we had to do all the tricky science bit first.

- [Nikolai] You mentioned how AI might, AI couldn't replace maybe like 30 years of a doctor's experience, for instance. I'm wondering what you think the sort of long term future is. Do you think, what are your general thoughts on AI in healthcare, and it augmenting decision making by physicians, by doctors?

And perhaps even taking over on some tasks or critical decisions.

- [Richard] I think it would take over the simpler measurement tasks, so there will be no need, you wouldn't need a nurse to measure someone's waist with a tape measure again. So that takes away that task and makes the nurse more available to deal with the patient. I think in terms of AI and the advances in this, being able to take away a doctor's judgment, I don't think that's either needed or necessary. And in fact, I don't think many patients would ultimately want that. I think, as I said before, it gives, it will give doctors a more informed starting point of that patient to be able to make a better informed judgment. I still think there will be a need for that human interaction, and I think most people want to talk to my doctor. I don't think that's going to go away. Where I think this will help tremendously is if you've got, say, a doctor who's just moved into a new town, doesn't know anybody, gets his list of patients, and they'll be able to look at the AI generated dataset on their PC or whatever equipment is available in 30 years time or whatever it is and then be able to get an idea of what that patient physically looks like and their history of measurements before they walk in the door. And that's quite a powerful tool for them to have. There's lots of talk about AI creating virtual doctors in the future and that kind of thing.

I'm not sure. I think a lot of, there's a very common thing that we all have as human beings, and that is that we all want to live as long as possible. And we will try and find any way that we can possibly do that. And so I don't think that desire is ever going to go away. So there will be people, whether they're my age or slightly younger, who will be worried about a certain symptom that might mean more than they think it does, and they want to have it checked out.

And so I still think that human interaction will be needed. I think it may be needed just a bit less than now because the basic measurement stuff might be taken away first.

- [Ryan] The big thing to take away from this is these technologies can contribute to that longevity conversation and that ability for us to live longer because we have access to new information in more of a real-time setting to be able to identify things way earlier than maybe certain symptoms kicking in or your annual physical that you go to.

So all of this is very beneficial for us as humans to be able to better understand ourselves, better understand our bodies at any given time to help us live longer, help us be healthier, and enjoy ourselves and our lives way longer than we might when we maybe get older, but we have a lot of different health issues that could have been prevented had we known about it or been able to identify it through the use of these technologies and these AI models and things like that.

- [Richard] Yeah, absolutely. I really generally felt quite humbled when I read Adolphe Quetelet's book. And he invented BMI to have something to measure the human race. It was not meant to be measuring somebody for an operation or someone who's going to take pharmaceutical drugs and all the things that have happened since 1835.

And I'm also realistic that I genuinely, well, I know that we've created something better than BMI here, but it's not the full answer, there's other things that will come into play, there's technology will evolve, AI is just incredible in its power if you feed the right data into it to provide new insights, and I think that's a really amazing thing to be able to think about, but I don't know the answers to it.

I don't know if anybody does. We need to get those, get the AI to help us understand those more. And I've just had my DNA test done and that was fascinating. It showed up a lot of things I knew but also some things I didn't. And when I started then asking my family how come I've got 8% Scandinavian blood in me, they said, oh yeah, you're grand, your grandmother's grandfather came from Sweden, and this is the kind of insight that DNA has provided us that hopefully in some way, the physical attributes of the human body can be better understood by us having a better system than BMI and AI then adding to that.

- [Neil] Interesting underlying theme here. One of my big experiences with healthcare and data analytics, AI, all that, is there's so much siloing of the data. The lack of sharing, all that stuff, but also from the practice of like medicine, we've gotten very specialized, right? You go see an ENT for this, or you go see a pulmonologist for this, or call it that, we've bifurcated.

So everyone's looking like that, from that one X-ray, even though we might be able to look for 200 different findings, you have one person looking at one or two things, and that's what they're trained on. Do you think, Richard, that AI, it's a great source of analyzing data. It could also be not just a tool but the platform to reunify all this medical information together to get that holistic picture?

- [Richard] I think yes it can in part. The way that, what we've found, I can probably answer it a slightly different way. What we've done is a lot of research on the ethics of how people feel about sharing their information. What we have found out is that people have no problem with their data being shared with a healthcare professional if they feel that healthcare professional can help them with a judgment on that condition. And what I would hope is that that same data is available to all of those people who are specialists in a certain field. And if they find out something about that patient or some trend in the data, that they share that with their fellow professionals outside their particular expertise. I think that would be a fantastic thing to happen, and it doesn't demean from them as specialists, specialist people in that field, but I think it will be, it'll be, I get what you're saying, Neil, and I think it'd be a lovely thing to have if they shared what they found out in one field with another field to better understand us, I think.

- [Ryan] One of the things I wanted to ask you before we wrap up here is if we look at this from a community and society type level, how do you think AI is going to be used to help measure or better understand the general health of a population or an area of the world? What is, I'm sure this impacts resource allocation for governments and things like that, if they're able to better understand how certain groups of people are doing from a health perspective, but talk to us a little bit more about how that can be, how AI can play a role there.

- [Richard] We've been doing trials with the World Health Organization in Africa and in India. And they asked us to deploy the BVI tech in areas where they've never been able to collect data before. Because basically, all you need is a camera on a smartphone or a tablet or just a camera. So we've been doing it in four countries in Africa and in India.

And I think that has been fascinating. I can't say any more than that, but different populations have different health needs and often those, the underlying systems are quite primitive, under resourced, they don't have the money to go and buy huge great MRI machines or anything like that. So having something that can measure the population and individuals in equal measure is really valuable because obviously most people have got a smartphone even in the depths of Africa now, and I think that having, and then you've got the ethnicity issues, where you've got people of different ethnicities who just genetically have a different body shape, particularly those of Afro-Caribbean origin where much lower body weight but not necessarily bad weight.

A lot of Asian people tend to be shorter and thinner than say Caucasians, and you've got people in the sort of developed Northern hemisphere who have, again, different body shapes and different health needs as a nation. So I think understanding as a government what your population is and what it comprises of and why they are, it opens up the ability to assign resources to those most in need, which I think is a very big human wish that we all have. And I think that if this in some way allows people to make better decisions for that, then that's, that would be fantastic.

- [Ryan] Thank you so much for your time. Really appreciate it. And we're excited to get this out to our audience. It's a fascinating topic that I don't think a lot of us just as humans really have been aware of that's happening, and it's going to be great to shed some light on it, so thank you again for taking the time.

- [Richard] That's great. Thank you.