 Hello and welcome back to the Sports Biomechanics Lecture Series supported by the International Society of Biomechanics in Sports and sponsored by Vicon. I'm Stuart McElaine Naylor from the University of Suffolk, and today I'm joined by Dario Catzola, who is an associate professor in biomechanics at the University of Bath or Bath, depending whereabouts you're from. Dario is a member of the Rugby Science at Bath research group, and his research interests include the modelling and simulation of human movement to investigate injury mechanisms and to understand human function. So following on from one of the previous lectures in this series by Alex A. Tak, looking at rugby place kicking biomechanics. Dario is going to do a presentation today, looking at estimating spinal loading in rugby activities. So if anybody has any questions as we're going along, type it in the live chat on YouTube and then we'll get to those at the end of the presentation. But thanks very much for joining me Dario, and over to you. Well, thanks a lot Stuart. It's a great, very happy, you know, very happy and very excited to be here. It's a great initiative and you've done a great job in protecting this up and advertising as well very well. Well done. Thank you. So as you said today, our sport biomechanics application is about rugby really. And what I want to do today is give you guys a bit of an idea of what kind of methods you want to use when you want to estimate spinal loading in rugby activities. And also I'm going to go through what are the pros and cons of this kind of methods and what are the different ways to link them up in order to get out in a reliable, hopefully, and results and results that make sense as well. So before diving into the presentation, I would like to thank all my collaborators because one presented today is really a group work and and that our funders as well so mainly the RFU injured player foundation and the camera project funded by EPSSE. Right, so now introduce you to the main characters for this presentation that the rugby activities we're going to talk about in the next 45 minutes or so. And on the right, you've got an example of tackling rugby and, you know, you can decide to watch the video or not, if you prefer not to. And on the right hand side scrumming. I'm really to me, you know, these kind of activities are really fascinating. They kind of show great technical abilities and the physical ability of rugby players and, you know, if you think about tackling that it's a very high dynamic event. You know, collisions with forces going up to five kilonewtons. And if you want, you know, sort of an idea of what you're going to go where they're going to experience if you tackle so one is like a stopping a 200 kilo motorbike running at eight kilometers per hour against you so good luck with it. And, but this crumb is instead of a more like a group related kind of activities where there are very high forces as well. And, and the forces that they can get up to are about 14 kilonewtons as a whole kind of pack. Unfortunately from a from a biomechanics perspective or injury prevention perspective. We do analyze and study a lot those kinds of activities because of a quite alarming statistics. So, even though the number of catastrophic injuries and I'm talking about cervical spine catastrophic injuries in rugby is not very high. Unfortunately, the severity related those injuries is very high. And, and if you look at the catastrophic injuries only, where you can see is that 40% of those injuries are related to tacos as crumb. So as biomechanics or injury prevention researchers, what we want to do is to try to minimize as much as possible those injuries because that those injuries can also result in in tits or plegia or quadriplegia or parplegia. And, and having as you might understand for rugby players and from a societal level as well. But how do we do that. So, as you can see here in this slide, I included one of the quite the various injury prevention model you can find the literature and that this is from from Meklin 1992 and I'm not talking about it a lot because this is a series of seminars such such a great series that this kind of cycle has been already presented in series one by in week one sorry by Aliza Demsic. And what I want to show you is what you should do, if you want to answer a specific research question in injury about mechanics that might be related for example how to minimize touching injuries. And really to understand what's the magnitude of the injuries it's talking about. So it's more related to epidemiological studies. And understanding what's the extent in the injury as well. And the second step is understanding what are the causes of the injuries. And if you are an injury by your mechanism by your mechanism, this is essentially the step you want to work on. And that's a very important step, because when I when I talk about causes really what I want to say I want to say that I'm talking about injury mechanisms which are to me, the main and the most important kind of way to understand how to prevent those. And ideas from prevention must be related to this causes. And once we've got this kind of prevention we can test them out. From a biomechanics perspective, my view is that there are, you know, it's probably in rugby mainly there are very different studies that are trying to analyze this kind of causes. However, the studies are mainly either observational studies or studies based on video analysis or personal reports. And there are some quantitative kind of study where some kinematics and kinetic analysis have been shown. And these are the more quantitative but still are not as solid that question. And also, there's a recent focus on concussion, which is slightly different from focusing on survivors by an injury mechanism. So, what I believe is that really, if we really want to understand what are the injury mechanisms here, we need to do it in a really using integrated approach. And in the literature there is still an open debate between, you know, buckling or hyperflexion being the main mechanism I'm going to talk about later about those kinds of mechanisms. And, but still, the main problem to me is that all this kind of studies have in some sense failed in answering the question that I'm showing, you know, at the bottom of the slide. So, what's the link between the external load applied to the system and the internal load experienced by our joints? So in terms of the spine, what's the internal load experienced at the intervertebral joint level? And that's very much also related to the type of kinematics that you will see. So, if we can answer that question, then we can understand what are the injury mechanisms. Otherwise, it's very difficult to do so. So, now we got our research question, and we are very happy with it. To me, the second step is to understand what are the key variables that we need to measure in order to answer that question. And there is an injury by a mechanism. What I want to do if I want to understand what are the injury mechanisms, I want to be sure that I know enough about the external load applied to the system. And I know enough also about the internal load applied to the system. When I talk about the internal load, really, I'm talking about the load generated by the muscle forces here. But also the load related to the passive response, for example, the spine. So we're talking about ligaments and intervertebral discs. Also, if it's possible, I would like to know what's happening in terms of motion. So I would like to know how all the vertebrae are moving. Of course, we are not in ideal world and we are not. We cannot use dynamic MRI or X-ray while someone is tackling someone else. So that's probably very difficult to do, but that's in an ideal world of what I would like to have. And I strongly believe that if I did have this kind of information that I would probably be able to understand how the limits of our system are exceeded. And therefore, I would be able to link up what's happening, what's the link between the external load and the internal load as well. We have gone through the different variables now. And the idea is that knowing the variables we want to know are the methods that we can use in order to measure those. And I'm really a big fan about an integrated approach, which maybe means that I like to do many different things and nothing perfectly well, but a lot of things. And looking at this, you can see that there are three main kind of methods that I would like to highlight to answer those research questions. The first one is in vivo. So in vivo methods, methods are essentially methods that are based on analysis or something that is a living organism and is a whole living organism. And a rugby player is a living organism indeed. And what you can do with in vivo kind of measurement, you can measure what's happening actually in the real world. Maybe you can now go down to the, you know, very invasive measurement because it's not ethical possible, but you are able to measure and know kind of what's happening, you know, doing an actual tackle or an actual scrum. Consider also all the facts and all the, you know, the behavior of the rugby player as well. The second type of method is called in vitro in vitro methods in Latin in vitro means in glass. So essentially, in this case, rather than having a living organism, we've got only a part of the living organism that living is not longer. And what's happening here is that what you want to do is want to be able to focus your analysis on a part of that body and be able to replicate some conditions that are very close to the real world condition. And for example, here you can see that's there are three big spines that are loaded with the, you know, using an axle rake, or you can do some anthropometric testing devices here. And even though they are not this kind of methods not as realistic as the in vivo that will give you the opportunity to, for the first time, get some information about injurious events. And finally, what I truly love in my research is doing some in silico analysis in silico means that you're doing things on a computer essentially. And the thing that I really like is that if you are good enough to create a good simulation, you're able to integrate all the information coming from in vivo and information coming from in vitro in order to have a very reliable and realistic analysis and or simulation. But how do we link them up. So in the next two slides, I'm going to show you what I believe are the different pathways that you can follow depending on the type of data that you have, and the type of research question that you are trying to answer. Anyway, the first step is to get some information about what's happening in the real world. So to me, you know, it's very important to use all the data set coming from video analysis and analysis, you know, based on a statistic based on theological studies, in order to be able to cluster different type of for example, tackles or get all the information that we need to describe them. Once we got that, what you can do, you can try to replicate the same, the same kind of task in the in your lab and lab will be a more controlled environment and of course there's not going to be exactly the same tackle is not going to be as realistic as something that happened during a match or training session. But the lab will allow you to use very accurate kind of instrumentation so you can get very good kinematics very good ground reaction forces very good muscle activation for them. And this kind of information can be used as input for in silico analysis and you know I've said analysis so remember that said analysis here because in this case this kind of computer simulation are driven by data set. And what we can we what can we do with kind of in silico analysis. Well, you can run some procedures like inverse kinematics inverse dynamics. You can also know when the motion and the forces applied to the system, try to solve what was called the redundancy problem and and estimated what are the, the muscle activation that you know solve that motion really. And once you got the muscle activation you can also calculate the joint reaction forces. So if you know more about that. Again, this is great because Professor Balzopoulos, in a couple of weeks ago or more probably gave a fantastic presentation on this kind of the terminology that you want to use, and the different procedures that you would use to in this kind of analysis. So have a look to that if you wish. The other pathway is, sorry. Go back to that so in this case, if you do this, of course, we are trying to analyzing in an injurious event. So of course you can get a rubber player in the lab you can ask the rubber player to do many things, but not to get injured in your lab so you are, whether you study studying here you are studying non injurious event, really. So if you do this, which is more related to in vitro testing, then you know will allow to do something different. So the actual idea is the same you measure things here and then you run some analysis afterwards. But in this case, as you might already understood. So we are now analyzing something that isn't injurious. So you can run the same ID and that case or inverse kinematics and dynamics, and you can calculate general reaction forces. Although, you know, in this case, you need to know that there are no muscles unless you've got an ATD so an anthropometric testing device with actuators. So we have been talking about the analysis so we use like in vitro and vivo and in silica to run some analysis, and I was driven by in the by data really. But there is another approach you can have, which is mainly related to simulation. So the study part is the same. But the second step is actually very, very similar to the same you do some in vivo testing you do some analysis. And you guess an awesome information about, for example, what's the kinematics so the technique used, for example, doing a tackle, what's the muscle activation. And that's information that is very much related to a non injurious event, but it's very realistic you are measuring that in vivo on that player. At the same time, you can do something different you can say okay. I cannot do I cannot generate or replicate an injury with a with a with a living organism. So I hear Robbie player, and I can do with an ATD and I can replicate an injurious event. And what I get out is very, very important information, probably more reliable information, some extent about the load, which is an injurious load in this case, and about the tissue strain depending on type of, you know, in vitro analysis that you do. But the great thing now is that you can integrate this information and run simulation, rather than just analysis. So, yeah, now you're trying to replicating fully replicating an injury, having information that is coming from from the real world. So you have been like, technique coming from the rubber player muscle activation coming from the rubber player and load that is supposed to be an injurious load as well, or tissue strain as well. So just to clarify that. So usually when you run an analysis, it means that you just kind of analysis is related to more inverse simulation, where you know everything about the kinematics, you know everything, what the external kinematics, you know everything about the forces applied to it, and it's driven by some data. And it's mainly used to calculate parameters. So in inverse dynamics analysis is an inverse simulation indeed. A simulation, a proper simulation, if you want to call it like that, is a for war simulation, where really could be driven initially by by some data. So you need to tell the simulation, oh, you want to start having that kind of neck angle, you want to start having that muscle activation. But really, while you're creating this new motion, you generate new motion you generate something that hasn't happened in the real world, you're exploring different scenarios. And that's, and that's what you want to do. If you want to try to understand what are the different injury mechanisms in rugby for example. Right, so in hopefully that kind of help out to set the background and give you an idea of what are the methods and what are the different pathways that we can use, you can follow to answer the main research question and we have here which is what are the injury mechanisms. And now I'm going to show you, you know, examples of how well how we collected this kind of data on in relation to tackling and this crumbling. And what are the pros and cons relating to these kinds of methods, and then I'm going to show you the simulation results as well. So starting from in vivo. So that's an example of how we replicated a tackle in the lab. So we asked, I think about 16 rugby players to pop over to the bath, as I said, as you said before. And we use a motion capture system to collect the three dimensional motion of the of the both the player and the punchback that in that case was our kind of ball carrier. We also equipped with like some pressure sensors that we use to estimate forces, and we also use ground reaction forces to measure the. Sorry, I'm sorry some force play to measure ground reaction forces. At the same time we measure also muscle activation at the for this final muscle, instead of the mastoid trapezius to spinae. We also test different type of angles and because we wanted to know what was the effect of the direction of the tackle. And, and we also tested different. The, you know, the laterality so how the type of work the firm, you prefer performed using either the dominant or not dominant shoulder. This is just an example of how long and it takes and how many people you need to do something like that. And I can tell you that operator was only one other people just put in some double sided tape and extra things. But is a, it's good fun, it's quite, it's quite long, but it's a very good fun and you can see the, probably two hours preparation to have like a five or six cycles. All right, let's now understand what why is good or what was why the pros and cons of this kind of analysis. Well, as we said before, talking technique is usually, you know, this kind of allies will allow you to have to measure very good tackle technique with very accurately, and many relation to the head position had angles and triangles. And many actually reserve some Japanese research group published something that to me is very very important they also managed to ask, they managed to get rugby players to tackle with a bad technique, which is quite dangerous thing to do by that that's incredible valuable information that you want to have in order them to run simulation afterwards. But the very important thing is that you can measure muscle activation. As I said at the beginning, we want to know what's the internal load applied to the system and muscle muscles are generated internal load as well. And we are now what kind of studies. So that both in scumaging and tackling there is a pre activation level before the actual input so this kind of this dashed black line is the experimental MG and the input is happening after this point. And you can see there is an activation which goes up from between 40% to 7% of the maximum activation so they are pre activating before the input as of course you want to do before a collision. However, there are, there are some cons, let's say, so the forces only estimated using pressure sensors and depression sensor are calibrated and are okay in the sense that the magnitude is always could be using we found that the magnet can be quite high you get up to four to six kilonewtons which is not completely, you know, out of the ballpark picture you expect but really you need is not like having a load cell pressure sensors not as reliable. The good thing is that usually the shape of the curve that you get out is quite good which is again it's very important if you want to run simulation because at least if you've got the shape of the curve then you can play with the magnet. The other kind of method we use is relative to in vitro methods and you can what we did we use both on to put method testing devices. So dummy has here, which is very handy because you can measure very well forces and you can get your colleagues around after running around running after sorry a punch bag. And here you have some you can use also some animal specimens. I mentioned that before as well. These are three pigs by which are loaded with a scrimmage impact blow it. And in terms of the instruments you want to have you want to use kinematics motion capture system sorry, you want to you can use the IC which is digital image correlation. And you can also you want to use some load cells to, for example during the animal specimen related tests to measure the load the cranial and the codal load. Yeah. This is just an example mode is a little example of what we have done so we, we, you know, we became very good friend of the local butcher we went down we got some young spines and we gave them a walk essentially that's what we did. And, and, you know, from a technical perspective what is very important to highlight here is that using the DIC where you can do, you can get information about the strain that are happening at different vertebral level so I apologize because I believe that for the first time you see this video you don't understand what you're looking at and I can tell you that these are the different vertebrae and there are some intervertebral discs in between. And these kind of different colors will give you an idea of the amount of strain that you have during that impact as well. So for the first time here, we can replicate an injury, as you can see we did replicate some injuries. And we had some structural damage as well. And the structural damage that we saw actually was a lot of damage that was very similar to the typical injuries that you will, you will get on the rugby pitch unfortunately. Both for scrimmage and tackling the bilateral phase at this location in the lower cervical spine is quite common unfortunately, and we did see those injuries on those kind of big spines. So I think that's the first step to say, you know, that's the loads that the load we are applying is good, the injury that we see is very similar. Essentially we are replicating as early the injury mechanism that is there, but that's a potential you need to be sure about it, and to be sure about it you need to have to make another step. What we did in terms of anthropometric, so, you know, ATD or like dummy head analysis, we tried to analyze the effect of the head angle during a misdirected load during a tackle. You can, there are some values there, and we did different speeds, and the high speed, you know, the peak went up to 2.4 kN and low speed 1.4 kN, and the kind of head down position was, you know, shown the highest load as well. However, I'm not a big fan about dummy heads really, because even though you can get a very good replication of the injuries event, you can get a replication of the injuries event. You can very much measure force in a reliable way, and you can use this kind of information as input for computer simulation. There is a big question mark related to the biofidelity of this kind of testing devices, which actually is not great. And, you know, the range of motion is not essentially the range of motion you expect. They are validated for specific inputs, so depending on the one that you use you want to be sure what kind of validation is related to that dummy head. And the stiffness is not usually the stiffness you expect during a tackle. Also, you are seeing something that is a passive response. We also tried to do like a bit of a hybrid kind of approach where we did some in-vito and in-vivo analysis, and we measured essentially, this is a dummy head here guys, ATD, and then there are, there is a three people scrum. And, yeah, so what we did, we measured the forces that we're using is to enter some machine and the forces using a dummy head, whilst we measured the EMGs on the player. So the good thing is that we got some EMGs doing an actual, well, an actual scrum, a machine scrum trial, whilst we get some forces on the dummy head. And also in this case, we did see a pre-activation of the muscle activation during the scrum. So, now we got, I think, a very good idea of what you can measure in a lab and what hopefully you can believe it's good or not. But next step is then to understand how to use the information in silico analysis. And if you're interested in analyzing the injury mechanisms, as are kind of, you know, injury by your mechanism, what you want to do is to get some knowledge about three, to me, three main areas. The first one is about muscle freedom modeling. So, muscle freedom models are essentially made up of a, is a multi-body system. And, and this kind of bodies are rigid. So there is a rigid body assumption, which is very important to understand if you're doing injury, injury by mechanics. The good thing about MSK models, you get both, you know, you've got bones, you've got bones, you've got ligaments, you've got muscles, and there are also muscle models. That means that you can use these kind of muscle models to solve what we call the redundancy problem and get out some muscle activation or replicate the effect of the muscle in informal simulation. So, this is something that you want to look at. And the second step is about comfort modeling. The comfort model, to me, is a, essentially, well, it's been, comfort model has been used like for decades. And the problem is that it's very difficult to use them. It's very difficult to validate them. It's very difficult to get reliable comfort models. But to me, they are the future. And because they will give you the ability to estimate the comfort forces. So, there is great work at the moment being done by Kieran Simms and Conor McCarty in Trinity College in Dublin. And I think that could be very exciting because you're being integrated with forward dynamics as well. And then the other kind of method that you want to look at is finite element analysis. And finite element analysis are much more detailed model with respect to MSK models, and they're not recent models, but each kind of body is made up of different elements, which are interconnected by a specific equation and stiffness values. And the great thing is that you can get information about the stresses and the strain that are applied. There are, experience a different part of the vertebra or the integrative discs whilst you're running a simulation. The problem is that it's very difficult to set up a simulation using an FE model. And it takes a long time. We need to remember that we are dealing with impact. So the classic, you know, causes study simulation that you do in FE are not really going to work. And so I think the approach starting from MSK models and then get the inputs, use the inputs from MSK model to run some finite element analysis is the way forward. Right. So, but how do we, you know, if you want to do some physical analysis, what's the first step? The first step is, of course, to understand what kind of model you want to use. And if you if you start from a MSK model, you want to answer, you know, a few questions before diving into some simulations. For example, you want to know, what is the model complexity that you need? Do you need a whole body model, or do you need a model that has got is just, you know, kind of focus on the spine and the head. So how many degrees of freedom do you need? Do you need a generic specific model that will change a lot. For example, in relation to the initial parameters that you're using or the muscle paths that are there. And as you know, very well, changing the muscle path with respect to a joint center that will change also the momentum. And if you change the amount of the momentum, the kind of low dispute that that joint can be very different. So you need to be very careful with that. The thing that you want to think about when you created the model is the passive response of your model. So you need to model the kind of structure that are the passive structure that are included in the model. And, and these are mainly ligaments and intervertebral discs in this case. There's a very good example from Quo and Camarillo in 2019. And they decided to go, you know, with a very detailed kind of model with all the ligaments. I'm going to show that we use a slightly different kind of approach. And, and also you want to know what kind of actuators you want to model. So you can have a talk to the model or muscle to the model, and there's nothing wrong with the talk to the model it depends on your research question really. And this kind of actuators. So if you use a muscle to the model again, you need to be sure where you get information about your muscle paths. So yeah, so we decided to have, you know, what kind of that's what that was the first step and let's say that we want to have like a had a neck model only and potentially a subject specific one. And when you decide to do that, then the first question that you ask yourself is, is my model good enough. Usually the answer is no, and then you ask yourself, is my model good enough for input events, and the answer is no, no. So it's a, it's very, so if you, you know, choosing the right model and choosing right model for the right application is very important. And in this case, what you want to be sure about is that at least the passive, the passive response of your model is the one that you respect, where you have like, for example, a very high dynamic axial impact. And that means that the interval to our joints kind of behavior passive behavior should be replicated fully to do that. So essentially we step from rubber players to pigs again. And, and while we do, essentially we go again to our local butcher, and then we say we need other spine, and the spine in this case, though, I want to be sure that I'm replicating very well, I can imagine very well what's happening during the max load, and even using a motion capture system. So we use some clusters here, and other kind of markers, anatomical markers to, you know, to register the model as well. And we measure the kinematics and the load as well. So the idea is that, you know, if I can get information about the kinematics I can get information about the load, then I can try to, I can replicate I got my subject specific model here. What I can do, I can use a genetic algorithms like a running optimization and say, rather than having all the ligaments I use very simple spring dump the kind of models, and see what that this kind of models can replicate the passive response of my system. So that's what we did. And that's a short video about the test. So that's the motion capture system marker markers, and, and after that you will see the actual simulation where in pink you see the markers, the simulated mark and blue the experimental one. And exactly, they were like, on top of each other so our model is fully following the replicating the experiments and we're very happy because at this point, what we have, we have a set of this elastic parameters, they are really characterizing the design, doing such a high dynamic events with Pavlos published a paper about it, and that's more than also with Professor Richard Gill and Dr. Itziberetoni as well. What's the, you know, what's good or what's bad so what are the cons related to this approach. Really, the problem is that, okay, I mean, you might love your local boat and of course you love you. But the problem is that you're using still an animal specimen, and the material properties and the interpretive angles as you can see here can be slightly different from a human. And, I mean, that's, it's not a really big problem for pig spine, because pig spine are widely used for this kind of analysis. And this sometimes is better, I think it's much better actually to use a animal specimen rather than a cadaver, which is, you know, probably 70 years old, or women. And it's not very representative of a rugby player either. And also the other thing is that what we're doing here, we are doing that mainly with axial impacts, we are in the process of analyzing the other kind of data set we've got with the, you know, kind of impulse with an angle so you can see that those parameters can be used for that as well. But that's still a bit of question mark as well. The second step is then to create, we got to create the this our subspecific model about and we that's in this case, we are talking about the rugby player. So what we did we asked a rugby player to do an MRI scan with the three D volumetric MRI scan. And we put some fish oil markers only to mainly to register their position then doing motion capture system analysis. Pavlos investors did all the segmentation I think he still has nightmares about it. And we can get all the bones and muscle. Sorry, bones and muscles segmented here. Also, what we did we wanted to provide better muscle parts here and we created some wrapping surfaces in math lab. So a sphere for the center column of stars to Tori for the torpedoes and a cylinder for the spinous capitals as well. In terms of the model, this is the final model and the final model that what happened. Well, there was the final model and and with all the muscles and the wrapping surface that was talking about before. Also what we did we did estimate the mass maximum as a metric force using the MRIs and the muscle volume as well. And which ended up being of course much bigger than a normal kind of healthy person because we're talking about rugby players indeed. But the problem in muscle is not only related to the muscle parts is not only related to the muscle parameters you want to use in a model is also related to getting some information about how they are activated during an actual event. And this is very important because if you think about if you look at the at this kind of schematic here in again what the forces are, you know, playing a role in doing a ninja are the external forces, gravity, inertial forces, and then the muscle forces. If you want to then, you know, really measure what's happening during a muscle at the neck muscle level. So what you want to do is to use some fine wire EMG. Of course, if you think about a tackle is from that's something that is not an ideal, and it's very difficult to use fine wire doing sport biomechanics as well. But that has been done for function movements. So what we could do really was slightly different. So, again, we could get the information about the anatomy and the muscles there. We run some tests, and we use for only four EMGs to start the mustard and to kind of two pieces EMGs to get an idea of how this kind of big muscle groups were activated. And once you have that essentially got all your kinematics you go on your external load that you measure doing those kind of tests. And what you can do now you can, you know, you go back to your analysis you say I can run a inverse analysis. I can I can solve my redundancy problem which is usually solved using static optimization. So the termization is that you're using a mathematical a priori kind of, you know, approach in order to understand how different muscle are activated. And that kind of, it's not great when you're doing spinal biomechanics it doesn't work very well. And also when you're dealing with inputs when you've got this kind of high degree of activation high degree of co-construction. In fact, if you try to do it. There's a kind of axial rotation. And as you can see there is a very much on and off behavior the muscle that's running blue and red here, which means that you've got like activation that you know, essentially, with an on and off behavior, which is something that is not really physiological. And you want to find different kind of solution for your kind of analysis on talking as chronology. Indeed, I'm not going through this schematic maybe if you got some question I can do it later. But what you can do you can get some trials calibration trials some execution trials, you run your inverse kinematics dynamics and you run your muscle analysis. And what you do you can try to understand whether you're solving the redundancy problem using a static optimization, or more like neuromuscular related models, like EMG assistant models can provide different results. And the hypothesis of course is that if you use information from the EMG to calibrate and run an EMG assistant model, then your activation are more physiologically plausible. That's an example of the simulation that we are talking simulation. And, as you can see, the we just use a head and neck in this case we added the arms as well because the muscles are touched to the scapula. And therefore the moment I will have changed with our positions. And, and what we can see here is that we are with top the actual simulation just just before the impact as well, because after the input the EMG wasn't that reliable, as you might imagine. So, you might ask yourself, so what's the result so is the static opt good enough, or you need to use EMG assisted kind of models to get a more physiologically plausible kind of results. And you can see the graphs two graphs here showing the CCI which is the co-contraction index during flexion extension and lateral bending kind of motion. And was it what you can see here, and if you can, you know, compare it to a second kind of EMG assisted kind of result, the level of co-contraction, which you find in blue for the static optimization is very close to zero, which means that the static fails to replicate co-contraction during such activities. If you start using EMG assisted kind of models, this kind of level of co-contraction increases, and it's very close to the kind of solid black line which is the actual co-contraction level coming from the experimental data set. And it becomes closer to minus one here for example. This is even better when you use EMG assisted model, which is informed by some MRIs and the subject specific values. So, they kind of take a message here is that if you're trying to understand what the muscle activation and you want to understand what's the best way to make the muscle activation during inverse analysis, static optimization is not the way forward, mainly because it fails at estimating the co-contractions. And what you want to do is using a EMG assisted model and get information about your kind of subject using MRIs or other type of tests you might do. This is also visible using the, you're looking at the three different models here, so as you can see the number of red muscles is increasing going from the top to bottom, and that's a essentially equivalent to this kind of graph showing the co-contraction in this as well. So that's great work that Pavlos did at Griffith University in the Gold Coast, so apart from surfing did something else as well, and that was done with Claudio Pizzolato and David Lloyd as well. So we talk about the good things about it and in the meantime you can see a scrum related simulation here. The cons about this kind of approach are related to the fact that it's rather time consuming, and we had all the data for a single participant, so we are now in the process to respond to that kind of analysis to other participants, but at the moment that's where we have. Right, so we are very close to the final part of the lecture. And finally I can show you some forward and endless simulation, super excited really. So I need to drink some water as well, and for the excitement. And here you can see there is a two different simulation of actually the same, so two different views of the same simulation. On the left hand side is a suggested view. And the right hand side is more like a frontal view of a of an head first impact, and applied to under the vertices of the ad. So it's like what we call the cranial impact. And now I'm going to show you how we set it up and what are the results that, you know, we got from a different simulation that we run like at these ones. The idea here was to try to explore all the different kind of possibility that they might have in terms of how a rugby player might, what kind of technique a rugby player might kind of embrace before or during a tackle. And to do that we wanted to span across different type of neck angles, more precisely under 77 neck angles. And we wanted to have different loading condition so different. The letter conditions were related to both the point of application of the force. There were three cranial one so anterior cranial cranial anterior cranial, cranial central and the cranial posterior, and the four lateral one as well here. So anterior, lateral, medial posterior, middle anterior and lateral anterior. Also, we had two different loading rates. And these loading rates are coming for the test that we did with the anthropometric testing device or with the dummy head that I showed before. There are two different kind of loading rates one for high speed and low speed. I think there's a typo. Yeah, sorry guys that would be the other way around. A few words about the simulation pipeline here essentially what we did. We use as an input in vivo in virtual data, and we started to simulate at impact time for 50 millisecond, and then we run for word dynamics analysis and we get out and we got out some integrative of joint loads. Right, so what are the results what are the take a message is eventually we took it took like six years to get to this point. I mean, at least to me. Let's see what we got out. We're going to go out. Well, so the first take a message is like neck flexion is bad neck flexion is bad. So I'm going to show you some graphs and just in a few seconds. The idea is that what we saw that across all the simulation that we run. So neck flexion neck flexion extension so movement in the subject of planes at the time of the impact had the largest effect on that internal loading so both compressive loaded and sheer loading. And the main reason that happened is because if you think about, you know, your spine and more neutral position, and then you're spending more flex position. What's happening when you flex in the spine, you making your spine more, your vertebra more aligned, making your vertebra more aligned that your spine become much stiffer. And if you then apply a force on a very stiff kind of a system what's happening is the energy is going through it. Three total and it's not dissipated. For example, using motion. So, what it means in terms of data. This is a quite busy slide so don't worry if you don't understand it. I don't understand it either. The thing is that what you want to look at is to look at the change in color. These here you got all the kind of reaction forces maximum reaction forces compressive reaction forces and to your procedure sheer reaction forces and across different integrative of joints. This is across different also at loading mainly cranial loading in this case. And what you can see the color is changing mainly in this direction. It's changing from a you, maybe I need to show the actual pointer is changing me in this direction. But this direction is flexion extension. So these changes are much more visible with respect to changes in in this direction and also this direction on vertical direction which are related to lateral bending and axial rotation. The same kind of thing that happens for the sheer loads as well as you can see here. A better graph to me so if you focus that kind of analysis only on the lower cervical spine. What you can see here is a if you think if you look at the milicranial kind of posterior so the red ones. The effect of the flexion extension angle is huge so they do guard when you don't very flex you got almost 3000 Newton as a compressive force. And while you are a standard you got here like about 1000 Newton so three times higher. And that happens also for the cranial central kind of loading. And it's not a different from the from the cranial anterior, but and if you look at the lateral bending mass rotation there is not that there isn't that kind of trend. So you can you can there is no much to look up in terms of what's happening in relation to this kind of motions. So we know that that's to me is very to make it more like a technique well kind of applied comment. If you're a coach and if you're a player, what you want to do when you go into a tackle, you don't want to flex your neck too much, otherwise you put yourself at risk. And of course, if you look at the, you know, values of the cranial cranial anterior kind of general reaction forces here, they are always high. And a kind of conditioning with which is very difficult to replicate you're not going to, unless you know, there are other kind of issues during the tackle but you're not going to go on purpose, hitting the other player with the cranial part of with anterior part of your head. There's a spoiler there, there's all about the second, the second kind of echo message, and that's essentially the most important kind of machine I think message of the lecture here, because really is the for the first time I think we managed to offer it to answer to a specific question relation to injury mechanism in rugby. Before explaining that I want to just want to clarify what are what what you know what are the diff the two main kind of mechanism that are in the analysis or you know what we want to analyze. And one is the hyperflexion so hyperflexion mechanism here is a mechanism related to a load applied to the posterior part of the skull. And finally, it's characterized by very high kind of flexion moment play on the spine, and the movement of the spine is clearly a flag a flexion movement for all the vertebra. A buckling mechanism is, is, is, but it's completely different. This kind of mechanism is characterized by merely compressive load to here. And, and the spine is got a behavior which is the first between the lower cervical spine and the upper cervical spine. And what's happening at the lowest of a spine are more flexed and the upper cervical spine more standing. But if an order of buckling there is a C kind of sorry there is a first order buckling where we see this kind of C shape, but there was a second order buckling when you see more as a shape and the vertebra as well. The first thing to look at is the, you know, if you if you look at the load, the type of load that we applied the type of load that you are, you know, likely to experience doing a tackle here is very much similar to the, you know, the direction of the load that you will have doing a buckling type of injury. So if you look at some simulation that we have, what you can see really is that you got a double, you know, a different behavior between the lower cervical spine the upper cervical spine where the lower cervical spine is, in fact, flexing and the upper cervical spine is bending as you expect to see during a buckling type of injury or buckling type of loading pattern, mainly. That was great to show, you know, what's happening during a simulation but I think very it's good practice to show also what's happening in terms of loading. And before, you know, coming to any kind of conclusion what you want to see is not just the actual motion of the head, which can, you know, can be, you know, can can lie to you in some extent because if you look for example just a video analysis data, you can find you can see some head movement which look more like a hyperflexion related mechanism but what you need to do is to integrate the motion of the vertebra with the loading. In fact, if you look at the loading, this is essentially this kind of graphs figures that got several graphs. If you look at the columns, you've got joint reaction forces in compression and shear anterior posterior shear and then flexion moments as well. And the rows will be the three different loading, the cranial loading condition that we have. So what you can see here again is that you got this kind of behavior of the of the vertebra so that I think you can you can look at is the how different the vertebra joints are behaving. And you can you can check their kind of general reaction forces values with respect to a injury threshold that we got from in detail and in people kind of studies from 90s and other studies here. So, looking at this, what you can see is that there is a clear loading, buckling loading pattern that is experienced by the cervical spine where you've got very high compressive load and this compressive load is very high, you know, across the board here, even of course for the cranial anterior as we said before, but for flex position is very high also for the cranial posterior and the shear forces are changing across the different reversible level, as you expect during a buckling, bucking loading mechanism. And, and the flexion moment is not that high. So to have a hyper flexion really you should have a very high flexion moment. Even though so what we if you see those kind of values you can say well they're very close to the limit but they're not actually over the limit and so it's everybody is, you know, is that an injury or not. That's a very difficult answer to, you know, that's a different question to answer and actually nobody knows and what we believe is important to this kind of highlight is that when, what we are assimilating here is the head first impact on doing a tackle. And what we believe is happening is that the, after the first impact, the cervical spine is buckling, is buckling in this way and is buckling and it gets to a new equilibrium point and that equilibrium point though, is a equilibrium point with a high loading, compressive loading shear forces which is classical buckling mechanisms. And after that, if you add on top some forces on the spine that would cause them I end up in different type of injuries. But we do believe that all these injuries are driven by this kind of bucking mechanisms. I have a more applied kind of comment. I think this kind of these results show the importance of technique. So I do believe that, as you can see, is much more important to have a very good tackling technique and be sure that you are not flexing too much your head and you are putting the head in the right place rather than for example then going through like strengthening program for the muscles which can also create some issues because you know the higher is the muscle force the higher is the compressive force that your spine as well. This is the last slide so apologies it was a bit longer than expected. And I would like just to give you three kind of take home messages of, you know, things that I believe are very important. In sport biomechanics, mainly in rugby activities as well. What you want to do is to be sure that you're using experimental data coming from different sources to then set up some computer simulation. This is key to have like reliable data. This is key to have like data that could be shown to, you know, coaches athletes that have, you know, a real application in the real world. One is if you're doing some modeling guys less is more. So if you use a very complex model is very much is very difficult to validate it is very difficult to verify it is very difficult to have control on it so and we have got so many kind of different inputs and so many things that can go wrong. You want to have a model that you can control very well. Finally, I think I would like to think about that we can we kind of demonstrate that in silica approaches can be used to inform the design of, you know, of intervention and ensure translation to real world decision making processes. At least they should be used to help out this kind of decision making processes rather than going through the testing rather than testing intervention directly on players. And that's it. I apologize. I was I was a bit late. Brilliant. Thanks, Dario. And yeah, I think that's great. I really like the thorough going through the integrated approach in vivo in vitro in silico. I think you said early on that you thought you might be a jack of all trades but master of none. I think after the last hour, it's definitely clear you're a master of all three. Really, really impressive. And I like kind of finishing on especially that point number two when it comes to modeling that less is definitely more in a lot of situations really. Yes, if anybody watching has got any questions type them in the chat and then we'll go through those in a moment. I think you've got your last slide. Brilliant. Thank you. So yes, if you think if you're interested in rugby then one of the previous talks by Alex ATAC was on place kicking biomechanics that you might find really interesting. But otherwise, kind of next week we've got Todd Pataki comparing discrete and continuous data analysis methods which is a really important topic in biomechanics generally. And the week after over on the ISPS YouTube channel, I believe I heard there's over 60 presentations going up. So they're all I think on average around about 10 minutes. But in place of the conference that would have been happening at this time of year, there'll be 60 conference presentations on a range of different sports biomechanics topics. So go and have a look at those. So question time. And I think, you know, the main kind of first one for me, Dario was just what's next. So I guess, what does the future hold for this area of estimating spinal load in rugby. Yeah, you know, that's a great question. There are many things that we need to improve really something that really probably haven't touched upon very much because for the sake of time is that, you know, the simulation that they have for our simulation. Can I go back with the slide that to show something as well. Yeah, go for it. You know, here we essentially use some vectors apply applying force to the to the to the head and that's a limitation to it because actually when you are tackling someone that kind of vector might change the point of application and might change magnitude with respect to what the other players doing. So to me the actual future is related to using content models integrated content reliable content models with forward dynamic simulations, and I'm lucky enough to to you know to collaborate with Kiran Simpson, Connor McCarthy from the Trinity College in Dublin and they've been doing great work on estimating the content parameters of content models using my demo which is a different type of model that usually it is usually using pedestrian accident kind of research. And I think this model can be passive or active, I believe that this one was mainly passive and they set up some for dynamic simulation, using the data that our, you know, we kind of collected our using the punching bag and the, and the rugby players, as I shown before, and they managed to actually replicate a very, very close behavioral of both the punching bag and the and the player doing that kind of impact so if you believe that there's a forward dynamic simulation, and the, you know, the, I think the why markers are the experimental one and the red one was going through is the, the simulated you get you get very close kind of behavioral both the player and and and the punch back so in a deal what you can if you got if we I can use this kind of content parameters to then simulate that to run the simulation I showed you before using content models and therefore having a much more realistic I think force generation that is changing depending on what for example that the ball carrier is doing best to me is the is the is the is the main one is a future kind of challenges. Thanks Daria. So next questions come through from Josh Baxter, who says really impressive work. Do you think your approach scales with athlete level. So do children tackle proportionate to their anatomy and their strength, compared to college and pro athletes, and how might that change your modeling or conclusions. Josh, great question, as usual. Yeah, so short answer is, I don't know. What I believe is that probably my answer would be no it doesn't scale directly you the forces are very different actually kind of youth rugby. It's got different rules as well so the you what you need to do is to be sure that you collect some experimental data, at least some video data to be sure that you are replicating, you know, the loading and the energy dissipation and the energy that is related to what kind of youth kind of rugby tackle or scrum. And, and the information about the, you know, muscle strength and you know you can play around with the muscle activation maybe but I think mainly you want you want to know more about the load and what's happened so it's not I think it's very much non linear what we see and can can change a lot depending on the on the players and the technique. Okay, thank you so you mentioned rules at one point in that that kind of leads me on to another thing I was thinking about. I'm not an expert in rugby by any means but I know there have been recent rule changes around tackling kind of what effect, could that have on any of this work and I guess what effect does does your work have on possible rule changes as well. If that makes sense. Thanks mate I tried to avoid that question as much as I could. No no yeah so I think, I think that's a great point and so as you might know, a new kind of tackling height rule has been trialled during the championship rugby and, and, and actually there was because the idea is that we wanted to, there was try mainly to minimize the concussion risk during tackling. And the idea with what they tried to do is to decrease the tackle height and what they saw a centre is that actually the bull carrier when they really knew that the tackle couldn't could have been performed very, but should have been performed very low kind of height, they start to crouch. So at that point the, the likelihood of getting cars got even greater because they had to have input would like even closer being the two players very close and crouching down that in that way. So, I think that you can, I think is, I think you can use this kind of simulation to understand what are the, the highest kind of guys risks related to tackle height and to do that though you need to have a content model so that's very much linked to what I said before, but if you have a content model and if you can model the interaction between the head and the actual thorax of the, of the other play of the bull carrier, then you can then simulate, you know, very specific conditions like the pocketing of the head, essentially, or you know how the, the, the, the load is transmitted, depending on the different positions. And yeah, so I believe, to me, that's, you know, in cynical kind of analysis, well it could at least be used to explore different hypotheses and then, then this kind of information can be used to inform different scenarios. So, yes, I hope we're going to do it. Great. I look forward to hearing more about that, but I think that's another really interesting example actually of how you change one parameter. And then everything else changes. So you might think changing one thing makes it safer, but then people adapt to those different constraints by changing their technique and suddenly something else is now the risk factor and then you've got to address that. Yeah, it's really interesting case study really there. Another question. So, could your model be used to evaluate strains on the skull, which dissipates as part of the impact? Very good question. And actually it's something I would like to clarify as well. No, so the answer is no. So we're using a MSK model and you cannot calculate any strain on the skull because it's an MSK model, it's a rigid body model. So what we, so when you use this kind of models, what you need to know is to when to stop the simulation. So if you've got a simulation that is generating deformation onto the skull, then the skull or other kind of structure, you need to essentially go back and rerun the simulation up to the point at which the simulation you think is reliable. In an ideal world that's, sorry, and we do that mainly because using MSK models is much easier than using a few models and that's there is a direct link between the in vivo in between data that you collect and you can set it up very easily. In an ideal world, actually, well, we started to do as well with other colleagues from my kinds of departments here at Bath, so Sabine Gadutzi, Wichigil and Bruno Hernandez, is to have a final element model as well. And what you can do, you can set the initial condition of the FE model, getting information from the simulation that you run with the MSK. So you don't need to run the full simulation with the FE model, but you run the simulation up to the point in which the MSK model is good enough, and then after that you run a FE model simulation. And in this case, if you want to get some strain or stresses kind of related type of analysis, that's what you're supposed to do. So it's taking your integrated approach even further. Yes, even further. Another type of model into the integrated approach. Yes. Okay, so yeah, last thing really for me is why pig? So was there a particular reason why you went for pig spines? Yeah, I asked that question to my close friend, Tim Hosgrove, the first time we did that kind of test and, you know, it's about, so there are many kind of studying the literature to mainly injury biomechanics, more kind of in vitro biomechanics, showing that the material properties of the vertebra and the vertebra discs of the pigs are quite similar to the one of the humans. And you can also use a vine kind of vertebra as well, so because sheep and cows as well. But it depends also on the availability. So and our local butcher have more like pigs than anything else now. And so I think, I think that's that's the reason why we use pigs. And the only problem with pigs that you need to be sure that the spinal curvature is not to dissimilar with to the human one. So if you've got like a specimen with a very high curvature, then it's not great because when you run tests, it's going to be very different from the human spine. Okay, interesting. Thanks for that. So yeah, just another huge thank you for the presentation and for the Q&A at the end. If anybody watches this back, say tomorrow or next week at any point, and they've got any questions. Is there a best way of getting in touch or asking any. Yeah, I mean, I'm very happy to do so you can drop me an email there is a my email address there. My slides, you can use or use my own Twitter as well. So be kind if you say something on Twitter with me, but I happy to answer questions there. Excellent. Thanks Daria. Thank you.