STEMology s1e26

[00:00:00] Sophie: Welcome to episode 26 of STEMology,

[00:00:03] David: A podcast sharing some of the interesting fun, and sometimes just patently bizarre news in science, technology, engineering, and maths.

[00:00:10] Sophie: Your hosts are Dr. David Farmer and Dr. Sophie Calabretto

[00:00:13] David: In today’s episode of STEMology. We’ll be chatting about hyper queen hummingbirds, tardigrade legs,

[00:00:19] Sophie: neonatal face heartbeats and drone delivered defibs.

Hyper queen Hummingbirds

[00:00:24] Sophie: Dragged up hummingbirds. Dave, did you know?

[00:00:37] David: Dragged up hummingbirds.

[00:00:40] Sophie: yeah,

[00:00:40] So this Dave has come out of Cornell University in New York and apparently it’s something weird is happening with the, It’s the white-necked Jacobin hummingbird. and what they’ve found is that over a quarter of females have the same brightly colored ornamentation as male. So they’ve got flashy, flashy wings, and heads and bodies in bright [00:01:00] colors just like the boys. But normally they shouldn’t. And the question is why, and it turns out that men are jerks and they love to body slam.

[00:01:09] David: They love to body slam. So yes. So the brightly colored birds helped them avoid aggressive male behaviors during feeding, including pecking, which is self-explanatory and body slamming, which is presumably just them banging into one another, as opposed to like the wrestling move.

[00:01:23] Sophie: Yeah. Like, I don’t know if their little wings are agile enough to pick up another bird and then just like slam it to the ground.

[00:01:30] David: So basically first author Jay Falk so this is a story about birds and the guy’s name is Jay. You can’t make this stuff up, but he had an aha moment when he realized that all of the juvenile females of the Jacobin hummingbird species had showy colors like the males. And usually the opposite is true.

[00:01:49] Usually if the juveniles are all one color, they’re the same color as the females. So they presumably I don’t know, like ducks blend into the foliage or whatever.

[00:01:59] Sophie: Yeah, and [00:02:00] then it’s something like also, normally if you have different looking male and female adults, often the juveniles already are looking different at at juvenile age. And then it’s like, this is a bit weird. So the idea is, as we’ve said that the male white-necked Jacobin hummingbird is known to have bright and flashy colors, iridescent blue heads, bright white tails and white bellies.

[00:02:20] Whereas this, and I think this is a main description because these are all the colors that I wear, but female Jacobins, usually a drabber in comparison and they wear muted colors. I don’t wear green, so muted, green, gray, black. And I was like, oh, everything I wear is gray and black. I’m a drab lady hummingbird

[00:02:39] David: No, you’re just a heterochromic lovely person.

[00:02:42] Sophie: Aw, thanks Dave. But yeah, so what they did is they went and they captured 436 white-necked Jacobins from Gamboa in Panama.

[00:02:50] David: That sounds like a pain in the arse. Like that sounds like a lot of work to capture 436 hummingbirds..

[00:02:55] Sophie: Do you think they just had a really big net and were quite fortuitous? Like,

[00:02:58] David: A really big net [00:03:00] and a really big flower. I just, the really big net above the really big flower and, Yeah 436 not incredibly intelligent hummingbirds, I guess.

[00:03:07] Sophie: No. Exactly. And then, and yeah. And so they, um, based on their plumage, they gave them a score basically in terms of their coloring. So as you said, you’ve got heterochromatic, which is essentially, I would say it looks like a girl is the way so it’s drab. Heterochromatic is drab in this case.

[00:03:22] Androchromatic, which is normally, looks like a boy, they’re flashy ones. And then you’ve got your mixed plumage types. So apparently 74.3% of the birds they captured were androchromatic 24.5% were heterochromatic, and then 1.6% were mixed plumage. But as you said, all juveniles at this androchromatic thing. And what they found was that all males will always androchromatic, but 28.6% of the females were also androchromatic.

[00:03:50] So this is this odd thing where we don’t have our drab heterochromatin females. We have some flashy females.

[00:03:56] David: Yes. We have some flashy females. So one of the hypotheses that [00:04:00] they set out to test here was, Well, is this a sexual selection thing? So the reason that the males are flashy colors is that’s a sexual selection thing. So basically it’s, the females will prefer the flashier males because having flashy plumage is somehow you know, associated with being healthy or having genetic diversity or something desirable for the females,

[00:04:20] Sophie: Well, I mean, Dave, as a drab woman. I am into a flashy man, if that helps.

[00:04:24] David: So women are into flashy men. And so they were saying, does this work the other way? Is there maybe a preference of the males for the flashier colors that would explain why these females are exhibiting the male colors? And they say no, because one of the things they observed with their 400 odd, captured, birds, was so they looked at the number of sexual advances and the number of instances of violence basically

[00:04:51] Sophie: With, like, with like stuffed bird mounts. I loved this. So it’s like, they basically, they got hummingbirds, real hummingbirds, and then they had sort of like to stuff to hummingbirds [00:05:00] that were different in some way. And they just looked at like what they did to which one? so very quickly the three different scenarios was you had a heterochrome female and, an androchrome male. So that is different sex, different plumage. You had a heterochrome female and androchrome female. I don’t think I’ve been saying that properly whole podcast so far, but it’s fine.

[00:05:20] David: I had never said these words before I read these words. In fact, the first time I said these words was talking to you now.

[00:05:25] Sophie: Okay. That’s good. Cause me uh, so that same-sex different plumage and then they had the androchrome female and the androchrome male. So different sex same plumage. And as you said, they basically just looked at like, when a real hummingbird is released at these two little mounts next to each other. What do the hummingbird actually do.

[00:05:41] David: Yes. And they found that the androchromic males would be quite aggressive towards the heterochromic females. So the male colored males would be nasty to the female colored females and including both just body slamming and also sexual behaviors, wherever those may be.

[00:05:58] Sophie: what was it they said that if there [00:06:00] was a heterochrome female mount, the first incidents of sexual behavior was directed towards the heterochrome female mount in a hundred percent of the trials. So like a hundred percent of the time, if the man was feeling like a little bit sexy, he’s getting sexy at the drab woman.

[00:06:15] David: Yes, is that why they called it a mount?

[00:06:17] Sophie: Oh gross. I didn’t even think about no, I let him know. I liked it. When, you know, when you kill an animal stuff it and , you mount it on the wall. That’s how

[00:06:25] David: So that’s

[00:06:25] why

[00:06:25] Sophie: because otherwise I just feel gross about this whole story. All

[00:06:28] David: So it’s ait’s a noun and not a verb. Okay.

[00:06:31] That’s good. I’m glad we’re clear on that. so they did that, so the male color males went for the female colored females. Now, when they looked at interestingly, when they looked at the now, correct me, if I’m wrong, the male colored males and the male colored females, there was an equal rate of violence between the two. So there was no bias in the direction of the aggressiveness. So both of them were aggressive against each other or sexy against each other.

[00:06:56] Sophie: Yeah. So I think that that’s the whole thing. So basically it’s just the drab [00:07:00] women who were being penalized in every way. So they’re either getting sexy advances or they’re getting violence at them, but they said that yet, if you had, the androchrome, male and female, the aggression was unbiased towards mount it literally made no. So yeah. Sex here does not actually play a role. It’s like the color that is changing the behavior, not the sex necessarily, which is interesting.

[00:07:24] David: Yes. So this is where it gets really interesting. You’re absolutely Right. because it becomes a question of, if it’s not sex driving the behavior, what is it, why do the females do this thing? And that’s kind of revealed by third group. So that was the androchromic females versus the heterochromic females.

[00:07:40] And what they found was that there was a bias and the male colored females were much more aggressive towards the female colored females then vice versa. So what that suggests is, so there won’t just be, sexual competition going on with these animals. There also be competition for things like resources.

[00:07:57] So they think that maybe by being male [00:08:00] colored, these female birds are reducing competition with other females for resources. So they’ve eliminated entire cohorts of hummingbirds with whom they’d usually have to compete. And therefore they’re more able to access resources.

[00:08:15] Sophie: Yeah. And so, um, the other thing that’s interesting about these birds, I think, and now correct me if I’m wrong is that, even if females have showy colors during the juvenile period, it’s never during the reproductive period. So basically if you want sexy times, you should be a drab woman, but in all other times, it actually is to your benefit to be boy colored in terms of, you know, resources and other things.

[00:08:38] So essentially. The females would only have showy colors precisely during the period that they weren’t looking for a mate. And then obviously, yeah, as you said, it means that they’re increasing their ability to compete for resources and other things, which is weird. Dave.

[00:08:52] David: It’s really weird. So it’s a really weird behavior and a really weird display of plumage by a really apparently weird [00:09:00] bird. And I think it’s a weird thing. The researcher said, he said something to the effect of, it’s really great that you don’t have to go looking for weird bird types to see new things in nature, but they got to go to Panama. Like that sounds like a pretty far-flung cool place.

[00:09:16] Sophie: Yeah. I don’t go to Panama

[00:09:17] David: I don’t get to go anywhere now

[00:09:19] Sophie: I mean, no one does really, but even in the times when we weren’t, all under house arrest, I also didn’t have the opportunity to go to Panama for anything.

[00:09:28] David: to look at hummingbirds or other.

[00:09:31] Sophie: Yeah. So, this is really cool. And apparently future studies, they hope to use these results of variation between female and male and colors and everything to understand how variation between males and females in other species may evolve.

[00:09:43] Because as we said, this is a little bit different, but maybe there are other species of birds and other animals where the women will be more successful if they dress like men, just in human society as well.

[00:09:53] David: Showy plumage and shoulder pads.

[00:09:55] Sophie: That’s right. If you want to be a lady CEO, just like a man and be aggressive, just [00:10:00] like a hummingbird,

[00:10:00]

[00:10:00] Sophie: Dave, I’ve got a question for you about our favorite little pals, the tardigrade.

Tardigrade legs

[00:10:14] Sophie: Why did they even evolve to walk?

[00:10:18] David: I don’t know. So this is some new research published in PNAS, hilarious, they’re named PNAS, which I was enjoying, which is

[00:10:24] Sophie: PNAS everyone playing at

[00:10:27] David: PNAS for everyone. Proceedings of the National Academy of Science, PNAS lol. And basically these researchers have looked at the way that tardigrades walk for the first time in a systematic way, because tardigrades not only do we not know what they are really, they’re classed in their own phylum. And to be clear, a phylum is a way of biologically classifying things that is very, very broad. So for example, arthropods is a phylum and that includes things as diverse as insect, scorpions, and crabs.

[00:10:57] So tardigrades have their own one of these because they’re so weird. We [00:11:00] don’t know what they are. So not only do we not know what they are, we don’t know why specifically they even have legs because they’re small and soft bodied and usually quick creatures that are small and soft bodies don’t bother to have legs.

[00:11:14] Sophie: So we don’t know like what, we don’t know why we don’t know who, where, and now we know how

[00:11:20] we’re knocking off some of these questions. So yeah. I thought this was a really cool, yeah. Cause we’ve talked about tardigrades before and we use them for all these reasons. Cause they’re like weird little, things that are very resilient to all the horrible things we do to them.

[00:11:32] But yeah, I never realized people hadn’t actually looked specifically at how they walk. this was a new analysis where they basically yeah they’ve got microscopes and they not only do they look at how they walked, they also measured things like, kinematics and intellect coordination, on different substrates.

[00:11:48] And as you said that they found out that basically the tardigrades walk like things, like insects and things that are over sort of 500,000 times their size, as opposed [00:12:00] to, as you said that, you know, if you have think of like little squishy things like roundworms, which are a similar size of body type, apparently they get around just by like thrashing and slithering.

[00:12:09] Whereas these guys have like eight little legs and they literally walk around and there’s, I don’t know if you went to the original like materials for press release. And they’ve got a little video of it walking. It just looks like a crocodile. It looks like a little crocodile with, too many legs and without teeth, just like walking along, and what they did.

[00:12:27] So originally they put them on standard polished glass. just to look at them and they found apparently because they couldn’t engage their claws because the surface was slippery. They were really bad at walking on it. So that’s when we enter what they made to substrate. So there was a soft, a poly acrylamide gel and they engineered two of them.

[00:12:45] And one of them had a stiffness of 50 kilopascals and they tested to 23 animals on that one. And then another one had a stiffness of 10 kilopascals and they tested 20 animals on that and they found. I really want to quickly quote something. So this is [00:13:00] by Jasmine Nirody, who was the leading author. And so when they were observing them, it’s my, and I’m sure you wrote it down too, Dave,

[00:13:06] David: I did

[00:13:07] Sophie: So yeah, we didn’t force them to do anything. Sometimes they would be really chill and just want to stroll around the substrate. Other times they see something they like and run towards. I love it. They’re just like, let’s just observe them, but

[00:13:18] they, yeah. So

[00:13:19] David: It wasn’t, it was not a case of Dance, tardigrades, dance!. It was

[00:13:23] Sophie: just hanging out on your substrate.

[00:13:24] Like we’ve made something that you can interact with with your claws. So we think you’re happy happy and just like do your thing.

[00:13:30] David: Beautiful. So, they found that at their most leisurely, they would lumber around around half a body length per second, which still seems reasonably quick. at full throttle, they would do two body lengths in the same amount of time. And you mentioned that they walk in a way that resembles insects 500,000 times their size.

[00:13:48] So. what they mean by that is that when animals speed up, some of them change the way they walk.

[00:13:53] So, So the example they give here is that if you think of a horse

[00:13:57] Sophie: Yeah, this is a good one.

[00:13:58] David: Yeah, So if a horse goes [00:14:00] from walking to trotting to galloping, those are all completely different methods of moving the legs in order to achieve a greater speed.

[00:14:07] and that’s interesting because it suggests that there are different neural circuits that govern those different behaviors and that you’re engaging them sequentially but tardigrades don’t do that. What tardigrades did was when they went from walking to running, they would just speed up the rate at which they move their legs in the same pattern.

[00:14:24] And which is interesting because it suggests that there’s just one neural circuit doing it that one thing, just doing it faster. And they kind of point to this as the key finding, because this suggests that because they walk in the same way as insects, like fruit flies, ants, and other segmented, scurrying and creatures, it may suggest a common evolutionary ancestor.

[00:14:46] Sophie: Yeah.

[00:14:47] David: And for these, which would suggest because one of the problems is that we don’t know what tardigrades are. It might suggest that they are something. And if we can show that they share an ancestor with something else, it means we can start to [00:15:00] group them better and understand them better.

[00:15:02] Sophie: and then I guess any of the other possibility is if there’s no ancestral connection, then basically these things have independently arrived at the same sort of running and walking strategies because maybe it’s just evolutionarily advantageous, but like it’s. Yeah, it’s really interesting.

[00:15:15] It’s like, why would these tiny little things that could just squish about literally walk, as I said, like eight legged crocodiles.

[00:15:22] David: Eight legged crocodiles, Not tiny bears. Apparently.

[00:15:26] Sophie: I think it’s just, cause they’re quite they’re short, there was like quite flat to the ground. So like the length of the legs to me looked a little more like a crocodile length leg than a bear length leg that I think that’s where the crocodile comparison came in in my head.

[00:15:39] But I’m, if you want to know what we’re talking about, hit to the show notes and watch that video. Cause it’s very cute it’s like a little gray microscope video of this little like scarring

[00:15:48] David: And they have legs, which allowed them to scurry. And we don’t know why still, but now we know how

Neonatal face detector

[00:15:53] Sophie: [00:16:00] So researchers have designed a computer vision system that can automatically detect a tiny baby’s face in a hospital bed and remotely monitor its vital signs from a digital camera with the same accuracy as an ECG. But Dave, I’ve got a few problems, but let’s, let’s get into this. And I can tell you what I think might be some issues

[00:16:22] David: So they made a baby detector that use artificial intelligence based software to detect human faces as is common in adults. But those algorithms are all geared towards people in normal situations and often not geared towards babies. So they develop a face recognition algorithm that works specifically for babies and specifically for babies that are preterm babies.

[00:16:45] And they were interested in not just taking, looking at the babies and saying there’s a baby face, but using that information to, extract important physiological information like heart rate and respiratory.

[00:16:57] Sophie: Yeah. So the idea is that, you know, you’ve got [00:17:00] these, so premature babies. It’s really important to be monitoring these babies all the time because they’ve come out early, which means often there are sort of health issues involved and these poor little things have sort of quite thin skin and they’re, you know, there’s a lack of development in certain areas.

[00:17:13] And so they’re going to be all tubed up. And if you’re using sort of adhesives and stuff to attach things to these babies, their skin is really fragile and you can rip the skin and it can cause infection and all this kind of things. So the whole idea is about how can we monitor these babies in real time, in a way that is like a little bit less obtrusive.

[00:17:29] And as you said, like, you know, if you’re going to use something that uses you know, videos to detect the babies, you need your AI to know the difference between like a baby and a tube. For example, like this is the baby, this is the tube. and so what they’ve done is they had a data set of seven videos of babies in the neonatal intensive care unit or the

[00:17:48] NICU at Flinders medical center, to reliably detect the skin tones and faces of these babies. and what they did is then they matched these vital sign readings to an [00:18:00] ECG to just basically work out if it worked but Dave there’s. So the first problem I had with this, and it just seems odd and tell me if I’m overreacting.

[00:18:06] So they were filming the baby.

[00:18:08] At the same time that they were filming the ECG, ECG is not digital. And do they not output data? And why do we have to have a video camera looking at an ECG and a video? I get why we need the video camera looking at the baby. Why wouldn’t you just take the data from the machine?

[00:18:23] David: this completely passed me by that they videotaped the ECG. they could be proprietary, so it might actually be difficult. It could be the, I don’t know, I’m speculating. I’m giving them the benefit of the doubt. Might be that they’re proprietary systems and it’s difficult to get the raw data.

[00:18:36] Sophie: Okay. Cause I was just thinking, I was like, this seems odd, but anyway, and, they just use like, they’re just some Nikon cameras to do this, which was great. and then, so what they did is they took 10 minute long videos of each infant, but the whole problem is that babies move. And for this thing to work at this stage, they needed stable babies.

[00:18:54] So they took 10 seconds snippets of these videos. And so for each infant, Five samples were [00:19:00] used, so essentially a 10 minutes of each baby, but they used 50 seconds of the babies. And then it was the AI that detects, which basically selects that region of interest being the baby, not the wires and everything else, so got, seven babies, they trained there, convolutional neural network with 473 images, which doesn’t seem a huge amount. And the whole idea behind this is that there are things that cameras can pick up that we can’t. Right. So they’ve used a color based method to extract raw cardiac signals and emotion based method for respiratory signals.

[00:19:36] David: Yeah.

[00:19:37] Sophie: Um,

[00:19:37] David: they, so basically that one. And when your heart beats, then the pressure increase, you know, you have a systolic and diastolic blood pressure. And so you have, when your blood pulses, that’s your systolic and it’s high blood pressure. And then blood goes into all your arteries.

[00:19:50] And basically in the face, there’ll be lots of arteries, lots of tiny, tiny arteries. And as that pulse wave hits those arteries, you get them filling with blood. And basically that [00:20:00] causes a very, very subtle color change in the baby’s face, which you can see with the camera, but not with the human eye.

[00:20:05] Sophie: My first question, Dave, is, would this only work for babies of certain skin tones? Because I just think about this in, in absolute terms. So say if I go for a run, I get very sweaty and red, you can see the flush of blood in my face. Right. But if someone had more melanin and more pigment in their skin, I don’t know if that’s necessarily easier to you know, you can’t really see it. you know, I guess the darker, the skin is that color variation would be less and less because, you know, it’s basically, you don’t see like a flush in super dark skin. And does this only work for yeah, so only because, you know, we have this issue with like a lot of the AI training things to look at faces and stuff. We already know that they’re super biased to race because they’re often just trained on white people. And then I thought about, I mean the real spiritual one sounds fine. Cause that one is literally based on movement, right? Like when you breathe, there are certain things of you that are moving.

[00:20:56] So if you train your AI well enough to detect like this are the [00:21:00] nostrils flaring or whatever, assuming that you use people of all different colors, like that one still seems fine to me. But the actual heartbeat when I was wondering, and I guess, you know, they only used seven children, and I tried to find information about those children, but like,

[00:21:13] David: or the training faces

[00:21:15] Sophie: Well, and that I couldn’t find either and yeah.

[00:21:17] It just referred to these big, big data set. yeah. So, I mean, the idea is really interesting. So the whole idea is the AI picks up The skin area that they’re looking at. and then they do spatial averaging for the various things. Then you’ve got signal decomposition,, spectral analysis, and band pass, filtering peak detection.

[00:21:33] And that essentially spits out a heart and respiratory rate. But for at least what they did, it seemed to work quite well for these seven cases.

[00:21:42] David: Yes. That was my understanding too. And they said something interesting. So you mentioned that they had to take data from the babies when the babies weren’t moving. And I was concerned about that as well because babies are alive and therefore they move sometimes, but they also may again giving them the benefit of the doubt [00:22:00] and again, not being entirely expert in neonatal clinical care.

[00:22:05] And they also mentioned that the most significant challenge they encountered in collecting the video information was the fickle and unstable nature of readings from the hospital monitor attached. So that would be the readings that are collected by conventional means.

[00:22:19] Sophie: Yeah, but is that like the video of the ECG?

[00:22:22] David: It’s the video of the ECG? but I don’t know if they mean that or if they mean that the readings coming from the equipment were actually a bit fickle anyway. because ECG, because it’s detecting electrical fields will also be influenced by the electrical fields influenced by muscle or can be, I don’t know if that’s a big problem, but again, if the babies move around, it may be that you don’t get great data anyway.

[00:22:43] Sophie: Yeah. So they said that there’s a close correlation between measure data and the reference data for both the heart and respiratory rates and there was a mean bias of 0.44 beats per minute and 0.71 breaths per minute, which is pretty good. So thinking over a minute, you’re getting this thing wrong by on [00:23:00] average about like 0.44 and 0.71. So you’re not even out by one a minutes. That sounds pretty good. And they list their sort of lower and upper limits of agreements. It’s looks like worst case for heartbeats, you get out by about five a minute and for breaths, it’s about six a minute. but yeah, I mean, they do actually flat at the end of the paper, they do flag all of the issues that, potentially things that they can or will try to address.

[00:23:24] Cause they’ve said and I didn’t understand exactly all of like the decomposition of the signals and everything without doing that was all way above my head. But they’ve said that apparently the techniques that they use produce noisy intrinsic mode function components, especially when the ensemble number is relatively low and that can lead to errors in the reconstructed signal.

[00:23:43] So I don’t know what that

[00:23:44] means, but they’re flagged that that’s, that’s an issue. and, in the future, to continuously monitor vital signs in the NICU they’re going to need more advanced signal processing techniques. but then there’s all that, those practical challenges that we talked about about camera movements, subject movement, and then even like the [00:24:00] illumination variations.

[00:24:01] So, you know, cause they, take an average of color changes what they’re looking at, but obviously that’s going to be effected by lights in the room and lights from the other machines and, you know, even, yeah, the cameras themselves and stuff, but like, it’s a nice idea, but I was just the first thing that happened when I read this, I went, what about the babies?

[00:24:16] That aren’t super white.

[00:24:19] David: Yeah. White and red. Yes, you’re absolutely right. So interesting attempt to do something interesting. Maybe not ready and maybe slightly problematic.

[00:24:28] Sophie: But, uh, we wait with bated breath.

[00:24:30] David: From the beginning of life to avoiding the end of it

Droned Defibs

[00:24:43] Sophie: Yeah, Dave. So there’s this interesting a story that’s come out of the Karolinska university hospital in Stockholm, and it’s essentially about delivering defibrillators to people via drone when they’ve had a heart attack in an attempt to like improve that hefty [00:25:00] mortality rate.

[00:25:00] Dave, I didn’t know this cardiac arrest caused one in five deaths in industrialized countries. I knew that was a thing. The mortality rate is 90%. That’s huge.

[00:25:10] David: Huge. It’s fatal without resuscitation and then some kind of electric shock. So you need some kind of external intervention every minute without treatment, decreases the chance of survival and survival is not increasing over the years. So it’s not something that’s getting better. and in fact, emergency medical response response times are apparently getting longer.

[00:25:30] So it may even in this situation where you need to reach people quickly.

[00:25:34] Sophie: Yeah,

[00:25:34] David: It may actually be getting worse. So, they’re looking, this is a feasibility study, a clinical feasibility study, looking at delivering automated external defibrillators, just outside the door of residential homes. Now I confused myself here because it said just outside the door of residential homes where most cardiac arrests occur. Why are these people having cardiac arrests on their doorsteps all the time? So I did a bunch of searching for it and I just got a bunch of , stories and anecdotes about [00:26:00] people being at death’s doorstep.

[00:26:01] Sophie: Oh,

[00:26:02] David: I find that I meant I finally realized that, Oh, it’s not the doorstep. That’s just the delivery site.

[00:26:06] They’re just having cardiac arrest at home, which makes a lot more sense.

[00:26:10] Sophie: cause I did the same thing, but I was like, is it because like they feel something is happening. They go to leave their house to seek medical attention and then they have, but no, I think, you’re right. I think it’s like they’re having them in their homes, but like you can’t.

[00:26:21] a drone, cannot penetrate a person’s house. And most of the window is open, which means that they’re going to drop them on the doorstep. Cause that’s like maybe like even more polite as well. I feel like Scandinavian countries quite polite.

[00:26:32] David: You’re absolutely right.

[00:26:33] Anyway.

[00:26:34] So, basically what usually happens is when a suspect to cardiac arrest occurs you know, hopefully a witness will call the emergency number, which is 1, 1, 2 in Sweden. If anyone’s wondering,

[00:26:44] Dispatch sends an alarm to the ambulances who drive to the scene, like it’s, it’s pretty standard. So in this study, in addition to that, they set up three drones at three different locations in Gothenburg, Sweden, and each with a five kilometer flight range. And if anyone’s interested, they were standard [00:27:00] issue DGI matrice 600 Pro Hexacopter drones. I just love that we live in a future where there’s like a standard kind of drone,

[00:27:07] Sophie: No that I wrote that down too. And I was like, if anyone wants to play at home, you just need to get one of those and attach a defib to it. But also Dave, to be part of this study, though, if you were going to pilot it, you actually have to do a remote pilot training, which consists of a one day general training in accordance with the Swedish transport agencies category 2 certificate.

[00:27:25] And then you have to do a five day internal study pilot training at Everdrone company.

[00:27:31] So if you have that drone, you’re there almost, you just need to do the training, especially with Everdrone

[00:27:37] David: Everdrone sounded like they would be the company that have drone for the longest

[00:27:41] Sophie: Yeah. Like the most amount of droning for the longest go and do some training with them. yeah, so I think the important thing to note here is that, as you said, it was in the Gothenburg city area, but it’s like airspace of the airport controlled airspace at the airport, because the whole idea is we can’t just be unleashing drones Willy nilly all over the place.

[00:27:59] So the idea [00:28:00] was yet when you call the ambulance. And you noted that this is cardiac arrest. Then they call in, to basically the airport and get like the drone dispatch center and get permission to unleash the drones because we don’t want, is an attempt to save someone’s life, you end up like crashing another plane,

[00:28:17] David: that’s, a very reasonable precaution. And then worthy addition to the study.

[00:28:21] Sophie: Yes.

[00:28:22] David: So, basically these drones they were equipped with, especially designed hardware and software, again from Everdrone , including a sense and avoid system. If anyone’s ever worried about the safety of drones, it’s never occurred to me.

[00:28:32] But if anyone is, they did have an emergency parachute which is pretty adorable. And they were fitted with a special winch system,

[00:28:39] which allows you to deliver the AEDs to the suspected cardiac arrests. So basically, these drones would arrive at the site. The drone pilot would kind of survey the site to make sure it was safe.

[00:28:50] And then it would winch down from an altitude of 30 meters, this automated external defibrillator, which beeps to let you know that it’s there. And again, very polite.

[00:28:59] Sophie: [00:29:00] yeah.

[00:29:00] David: It’s a beep. It’s not a siren.

[00:29:01] Sophie: No beeps politely for the bystander to go and retrieve it and then, perform some first aid. But I think what the interesting thing was like, so apparently the drones were automated, but at all stages they’ll be watched by the drone pilot, because obviously, as you said, this is a proof of concept as well.

[00:29:15] So we want to make sure we’re monitoring things and apparently they, so this is interesting. So in June to September in 2020, 14 cases of cardiac arrest qualified for inclusion, but 39 didn’t. And that’s because there are things like, so apparently the drones that they use can’t deal with like heavy wind and rain.

[00:29:32] Which is something that you can fix in a drone, but there are also issues of being in no fly zones or in sort of in you know, high rise building areas. so of the 14 that did qualify only 12 took off. And I don’t think they said why two of them couldn’t. And then of that 12 who took off, 11 of these drones successfully delivered AED, which is pretty like this 92%. That’s like pretty good delivery rate. and so the median flight distance was 3.1 kilometers, and the drones [00:30:00] arrived a median of nine meters from the victim. Which is not like 9 meteres is pretty small.

[00:30:04] Like that’s pretty good. And then apparently the drone arrived before the ambulance in 64% of the cases, the time benefit on average of one minute and 52 seconds, which as you said, like every minute that we’re not doing something about this cardiac arrest, we increasing the chance of death.

[00:30:18] David: A bit better than half the time people were administering the fibrillation before the paramedics got there. And to clarify like that doesn’t sound great, but like you say, every minute counts and also like They had to plot quite a complicated route to the house they had to reach because the optimal flight path would focus on spending a minimum proportion of time above populated areas, just in case the drone crashed. Presumably.

[00:30:44] So actually that is quite complicated. So to have had this positive result with all these constraints. It’s kind of good. And my favorite thing, I think about the paper and I think they’ve done something very understated and classy is they haven’t made any [00:31:00] comment at all on whether they improved survivability in these patients.

[00:31:03] Sophie: But do you know why that is Dave? I don’t know if you saw this sentence in the end of the paper, but they said, even though a drone delivered an AED before the arrival of the EMS in 64% of the cases, no drone delivered AED was used in the feasibility study, no analysis of data on AED use and experiences of bystanders onsite are therefore possible.

[00:31:23] So I don’t know if they didn’t tell the bystander that this was so, you know, cause my understanding is like the AED that they delivered could have been used. Like it’s not like they went, we’re just deploying boxes that are pretending to be AED to see if we can deliver them. So none of the defibs actually got used.

[00:31:39] And I don’t know if like it wasn’t clear that they could be used or. If you think about like, if you’ve never used a defib machine, like that might be a bit confronting if like your, one of your loved ones is dying and you’re like, am I going to make this better or worse by doing this thing? So, yeah, they didn’t actually use them.

[00:31:55] David: Okay. That explains why I thought they just hadn’t mentioned it because[00:32:00] they were being classic. Cause they didn’t have statistical power, but maybe they knew that was going to be the case. And that’s why

[00:32:04] Sophie: Yeah. So I don’t know if they need to, because you think they would, if you’re on the phone to, one, one, two, and they’ve said an ambulance is coming also a drone is coming. We’re coming with like a defib machine. Like, the older automated ones are pretty like, so I’ve done a first aid course.

[00:32:19] So I feel okay using a defib machine, but they’re pretty light. The instructions are good. Like you open it up and it’s like, put these things here, like press this thing here now. Like, but yeah. I don’t know. They don’t actually talk about why they weren’t use, but it was more that they weren’t.

[00:32:31] But the idea is if you know, they’re getting there earlier and people can use them, then that you’re adding, minutes on top of your potential saving people, timeline, which in a cardiac arrest where your heart literally stops. It’s you know, those minutes are quite important to get that pumping again..

[00:32:46] David: Yeah, I have to say if someone called me up, if I was like administering first aid to someone who which would have been an unskilled legs, I haven’t done a first aid course to

[00:32:53] Sophie: I mean, I did my like 10 years ago, so maybe don’t get me to do a CPR.

[00:32:57] David: called the medics and I was like, Hey, cardiac [00:33:00] arrest. And they were like, okay. an ambulance is coming also a drone.

[00:33:02] I’d be like, thanks very much for the help also. That’s awesome. Thank you. Goodbye.

[00:33:06] Sophie: Yeah. And then you’d be like, look at the drones and then the ambulance gets there and they’re like, when did they last say so you’re like, I’m sorry, I’m really busy to check out that drone to something.

[00:33:14] David: so very little clinical stuff in that clinical feasibility study, but pretty cool nonetheless.

[00:33:20] Sophie: And then they’ve said that, you know, this could be expanded to other medical scenarios, like the delivery of epinephrin to patients with anaphylaptic shock or delivery of glucose to diabetic patients with low blood sugar. Also, I don’t know if I buy the low blood sugar one, because just like most people I know with diabetes just always have, like jelly beans on their purse.

[00:33:39] David: Also would we be delivering jelly beans by drone? Because I don’t think that should be limited to medical situations. So I

[00:33:45] Sophie: my God. Can you imagine Uber eats, Deliveroo Door Dash or we’ve got a proposition for you? I’d like to have my food delivered by drone in the future. and then that way you don’t have to underpay or your delivery people [00:34:00] controversial.