Ciprian Crainiceanu presents at the Johns Hopkins Department of PM&R’s Grand Rounds on April 18, 2017.
For instance, they, they, they were trying to monitor easily using infrared cameras because if you patient with stroke, they don't move. Ok. We use, we use um we, we have started using cameras uh to, to look at care providers, uh the interaction and this is in kind of mental health. So we were, we were trying to, we just piloted one of those things and it seems to work. Uh we, we had several issues with cameras in people's phones, of course. But, but it seems to work out. Cool, cool. All right, I'm sure they will be. Hold on. I'm gonna sit and ok. Sounds good. Sounds good. I'm very, very happy to talk to you. I welcome everybody. So the really interesting hot topic for today. Um and a great speaker, the doctor. Uh um I'm sorry, practice beforehand, but I obviously failed um get uh his name right. But um doctor Channel is uh professor Professor of Bloomberg School of Public Health across space and um is uh professor biostatistics. There is an incredibly impressive uh curriculum. He's had over 100 and 20 publications. He's been uh leader of the study section for the N A for bios um among a number of other means, uh a group over there that looks at uh applying wearable technology, wearable, and implantable technology and using big data to analyze it. And basically this is a great topic for us and we have since we're so interested in activity and monitoring it and personally, I'm interested in it because I wear one of the fit. Um As I sure many of you do so with that, we'll hear some more about it. Thank you very much. Um Has anybody, has everybody seen this information on the screen? I was told to, to show it. Um First of all, thank you very much for inviting me here today. Um I hope that, you know, this is the beginning of a longer, at least discussion, maybe collaboration uh between your department and, and us. Um We started something called the smart group statistical methods for analysis of research technologies. And then uh one of the subgroups split into wearable and implantable technology. So wheat is a product of smart. Um If you want to learn more about us, uh smart stats is uh our website, we have a lot of information there um If you want to, to learn more about it. So who are we? And what do we do? We are four Bios Statistics Faculty. Um We have several R one grants, we have several additional contracts with Nian I Mh the Gates Foundation. Um We started working on wearable computing about six or seven years ago. So we have probably the largest number of studies um in wearable uh computing. We have 30 masters, about 30 collaborators at Jhu, hopefully more in the future. And we have started four small sister groups at Pen Columbia and C State and in Germany at Lud Maximilian. So what do we do? Uh we work on on wearable computing, Axon Mery GPS, hot rate, skin temperature, self report video cameras, E MA looks like there are a few chairs there. Um We also work on brain imaging but, but I'll talk less about this also, bios signal, ce GCGC blood pressure, um and so on and so forth. So here are some of the devices that we uh we used, I'm gonna go through, through them a little bit. So this is an older uh type of phone that was used for E MA. So, you know, when, when you ask a person you can ask them, well, what did you eat last year? Which is basically how we have a lot of our nutrition studies. Uh or you can ask them the question, how do you feel right now? Right? So they, they can get a uh a phone message and they can answer immediately. Well, I'm fine or I feel energized and this is uh it is believed that that this is a new type of uh of way, a better way of doing self report where you ask specifically what happens now or what happened in the previous hour. Um This is our other devices, some axel meters, active Watch Geneive Act. So when we work with, with our collaborators, one of the first question is what, what should we use? So we, we are familiar with, with many of these devices. This is the active heart device. This was using the Baltimore Long study of aging that I'm gonna talk about. Uh in this uh this is a device that was uh placed at the chest level and it monitored both the acceleration of the chest of the person and the heart rate continues for seven days. So this is the device that I'm gonna focus on. We, I'm gonna look more at the activity data though, we are just finishing with activity and heart together. We just finished the paper on that. Uh This is uh camera and more and more we are actually very interested in pairing what we call devices that do not see with devices that do see so that we get the additional information in terms of what you are measuring. I'm sure I'm going to show you in a moment exactly how this is. And it's a very, very powerful way for us humans to actually understand the data because these devices, they are kind of black boxes in a sense you put them on and you kind of have to trust them. But for somebody like me, it's very important to be able to monitor and see what's happening with a camera in the same time to look at the data and see the connection between the two. So these are the types of things that uh we kind of uh implemented. I think that other people also came up with these ideas. But it makes for, for better design of experiments whenever you want to do things. Um Also Z patch, this is an ECG monitor from Stanford. There are, there are other, many other things. Um And this is because we're now also doing plant technology uh together with Naresh Punjabi in uh uh um And this is actually what I have. This is an Abbot monitor is a Gluco glucometer. And both I and Urbane got implanted for a week because we wanted to know what our uh glucose levels are for uh for two weeks. He got another one in, in his belly and he looked at whether the two actually correspond to one another. Um uh We, we are pretty healthy from what I remember. Um But in general, I think it's, it's important for, for me to understand what the measurement is and to try to, to come up with, with a way to explain it to others as well. So we, we do other things here was trying to, to see whether walking on crutches differs from walking. Normally from the point of view of the data, we all know if you see somebody on crutches moving or you see that that's different from moving normally. But does that result in a signal change that is measurable at the level of the device? It is not intuitively, it should be right? Because somebody is limping, their, their, their uh leg moves in a different way. But is it true that in fact, there is a, an actual change in the signal in the wave form in, in the signal and this is something we, we like to try. So let me show you um a movie that we made and this is, it's using a camera that sits on my chest. Um And here on the left side, I have an axel meter. It's, it's called the gene active. Um I also have a second one on my right wrist and what you see here, this is actually the data. So I'm opening up the black box for you. OK? I'm just, I just told you this is a black box and it generates some signal. What is it? It's basically acceleration in terms of GG being the earth gravitational units. OK. So how many GS do you think you develop when you walk, when you walk around? How many GS would you say? Sorry, 1 to 21 to 2 GS? So, so typically less than that, depending on, on how hard you strike or uh but you know, close to 0.5 to, to one G. Uh These are three axes because the accelerator has three axes back forward, left, right up down. And you will see how this is moving as I, as I start. Um, the movie. So the top panel is just my left wrist. So this is that, of course, the right wrist is not yet there. And I'm not doing anything special. Of course, the movie is like 12 hours and I, they told me that I can only speak for 15 minutes. So I could, I could have shown you the whole thing. It's really not interesting. Um But I just wanted to you to see a little bit what, what is happening? So these are the data it moves as. So it's just a nice thing to kind of look at for me because I start to understand what's happening. So there are some things here. First of all, the time series seems to to shift up and down. The blue one was up, then it's down, then the green one is up. So it kind of makes you wonder what's happening. So what is happening is that as I move my wrist around, what used to be the gravity pulling in one direction is now pulling in a different direction. So that's what's changing the um uh what is happening. The other thing is that if I only showed you this data and I only show you this time series, you would have no idea what I'm doing here. So now I put my second axel meter and this started moving. I'm not doing too much. So it's just a little bit. Uh And I'm moving my hands around. I'm not doing basically more than what you do when you sit in front of your computer. You know, I kind of look for things. Nothing very special, but I'm, I'm pretty active from the point of view of the accel meter. If I had an acceler on my hip, it would be zero, it wouldn't move at all. So it gives you an idea about, you know, the various things. Uh what this is measuring is the acceleration of my wrist and, and that's, that's a pretty important concept because uh most people tend to think that it's activity. No, it's acceleration of the particular. And now I'm just trying to move a little bit harder and I generate uh somewhere close to four or five GS. And at the time, we didn't know how to uh synchronize the image with the time series. That's why I was doing that because now we have better, better ways of much better ways of doing that. But anyway, so I just want to show you a little bit about what's happening behind the scenes and what is kind of the raw level of the data. What what do you see when uh with these type of devices? So just to get you uh a little bit more in terms of uh give you a better idea in terms of what it is that he is measuring, this is what you have seen on in the movie, which is data that is sampled very, very frequently, maybe 30 Hertz, 30 times per second, maybe 100 times per second. And this is data for uh five days. So this is basically the person being active during the day then sleeping. It's pretty clear when the the person is sleeping. So the device picks up that very, very easily. Um Now I I took one data and I blew it up a little bit and you see there is more detail in it as, as you go. And then I, I just looked at the six minute interval and there is detail there as well. Just want to show you just how many layers of detail can there be uh in there. However, in practice, uh you know, uh the fitbits and the devices that we are using typically using, you never get to see this data. OK? This data you never get to see is just up immediately transformed into numbers and the numbers are for a minute, which is somewhere in the order of a few thousands of uh observations. It's giving you a number which is called the activity count, sometimes the number of steps and this is what a time series, a number of steps per minute looks like. So you have the activity, zero steps and so on. And so forth. So this is the type of data that you would get from, from a device. Of course, most devices do not give you that. They give you a summary, they, they take the data, they take this data and say your total activity count today was 9000 steps. Go get 1000 more because that's useful. That's basically the level that we are talking about. And that's why you can you start from millions and minutes of observation and you end up with, with one data point and maybe that's all you need. But as a statistician, I have to ask the question. Is there more information there? Is there a better way of thinking about the problem? Is there a better way for us to, to move forward? And sometimes it is sometimes it's not the total activity count is tends to be a pretty powerful thing to look at. But uh I'm very interested in what happens to the circadian rhythm. I'm very interested in what happens to the individuals. So I'm gonna show you several things that uh we have done. So I'm gonna go a little bit faster over this. The first thing is that we looked at, we did an experiment, an in lab experiment with Tom Glass. Um it was a with three subjects, 10 activities repeated two or three times. Um And this is what data look like. So when Tom came to me, he said I have this data and I I will tell you this is working. Can you recognize that this is walking, right? I mean, when you look at it, you immediately probably trust me that if this is working, this is also walking. But I I would like to ask you, why do you trust me if I tell that this is working? Why do you trust that this is working? It's oscillating kind of correlate steps, it's oscillating, right? So, but but more than that, this is this is also oscillating. Why is this oscillating? Working? And, and this is not working, everything is oscillating. I'm telling you everything is oscillating all the time with an el matter, different magnitudes. So, so you started to look at things into the magnitudes tend to be similar, right? I think that our brains are probably saying there are some patterns that look the same. But what do you mean by pattern? It's very easy for the human mind to just I see I've seen that before, but it's very hard for the computer to do the same thing. If you push this into the computer and says, find me when it's walking, it's not gonna do it, it's gonna give up. So we are very good at seeing this. Now, let me show you another. This is from the same study, but this is the person who was asked to stand up from a chair three times. Can you see the three times? I see some nodding So I guess we all see that the person is standing out from the chair three times. It's in gray. And again, you kind of believe that he's standing up from a chair because I said that this is standing up from a chair. Right? Again, why is it that this standing up from a chair, this standing up from a chair? Why do you believe? Why is it reasonable to think that this is also standing up from a chair looks the same looks similar. How would, how would you prove to yourself that looks the same? I would come with some scissors, right? I'm not gonna do it here. I'm not gonna cut this but I have some imaginary scissors and I'm gonna cut it here and I'm gonna try to move it around and see where it fits. Well, that is one way of identifying patterns. OK. So it's, that's, that's basically the idea. The idea is how do you identify patterns? You take small pieces of the time series and you move them around and look for other pieces that look like themselves and believe it or not. That's exactly how speech recognition is done when you ask Siri for something that's exactly the same, the same technique that they are using is based on, on what is called dictionary learning. So they cut the small words and they look for words that look similar. The problem that they had was with a, with accents like like my accent, right? Because this didn't look identical to this. An accent would, would, would make things different. The same thing with walking, just like with talking, there are accents for walking as well. And that's something that we think is, is very important, especially if we try to model and interpret working as individuals who are not healthy. So it could be that we come up with a specific experiment that we ask individuals to, to do a part experiment that may provide us with the data that characterizes their specific uh pathology uh in terms of, in terms of anyway. So in short, that's what we did. We said if you tell me this is walking, I'm gonna take two seconds of walking. I'm gonna cut it to small little pieces and then I'm, I'm just gonna call this a dictionary. So it's I like I have a book with words and every time when you give me another data, I'm gonna look for one of these words in my dictionary and if it fits that's walking. So that, that's uh the idea uh here um I will move away from, from this kind of machine learning technique, which is interesting. It's, but it's not directly answering health problems. I'm gonna move towards looking into what can we do with activity intensity? OK. What are the types of things that we have done that may be interesting to you uh may maybe suggest some, some things that could be done in the future. So this is, remember, I'm moving away from this very, very fine data, the micro level data and I'm moving towards the macro level data or the step counts, the step counts per minute, right? So if you want, you can think about it, you look at your favorite Fitbit or whatever you have. Every, every minute, you look, how many steps have I done this minute? So you'll get 1440 observations because there are 1440 minutes per day. And that would be the data. So that's exactly the data that I have here. It's spiky, it looks ugly. I don't know exactly what to do with it, but we'll try to, to understand it a little bit more. So how do I think about this data? This is here. You have three subjects. These are data for three subjects, subject one, subject, two, subject three and they are observed on Monday, Tuesday, Wednesday, Thursday and Friday. Ok. So this is midnight to midnight, midnight to midnight, midnight to midnight and so on and so forth. So the way one way to organize the data could be to just well average over all of this and say this is the number of steps per day. If you do that, of course, you, you lose all the information about what happened within the day when have they walked? So I like to just plot the data. Have a look at it, try to understand it and try to see what, what the problems uh could be, but it's not enough. We would like to do some in, we would like to extract some information out of this. So what I do, I make a really bad plot. So I take all of the observations for this subject and this is a healthy control H 67 and this is their data. It looks very messy. It's grayed out. You don't even realize that this is probably night. This is midnight to about 6 a.m. So what if I take the average at the same minute? Hm. This starts to look a little bit more like what I expect. So this is the average during the night. So the activity counts during the night are lower. It kind of tells that that probably the person either wo woke up or went to the bathroom around 6 15, right? And then they started their morning activities, they had some lower activity and so on and so forth. So this starts to resemble a little bit what you would remember about your day. Oh, I woke up around seven or I woke up around five. I did this. I did that. That sounds, that sounds reasonable on average because if you, if you look back to your, to your day, each day will look quite different from, from your average day. And then you can do one more step. You can average over all individuals who are between 60 70 years old and are healthy. And this gives you an average profile of activity for individuals who are healthy and are between 60 70. So that's, that's very nice. It starts to make a lot of sense. So it says that these individuals don't sleep at all very well around midnight. Uh Most of the activities during the day is kind of sustained. There are not a lot of ups and downs in, in the averages. Why is that important? Because we would like to compare that with somebody who's not healthy, say somebody who had a stroke or somebody who may have diabetes. We would like to see what does those, what do those carum patterns look like for those individuals? That's why that's what we are getting at. So, one of the studies we worked on was is the Baltimore National study of aging Conduct NI A and led by uh Luigi Fucci. So, uh the first thing that we did and and this is something that in particular has been very active uh is to kind of put the data in a standard format. This is one of the, the biggest problems with, with the data uh of this time is that everybody has a different format and it's, it's, it's a pretty big mess when it comes to us. But uh we came up with a simple analytic format. We can show you exactly how that is done. And the good thing about it is that once you have the data in that format, things are way, way faster in terms of the analysis for, for a lot of us. So I would say that maybe 80% of the work is to take the data from the device to an alive format. So I I could tell you, oh, you know what we do is so much work. And no, we have solved that last the last mile problem. The real problem is to, to get the data and put it in a format that is analytic. OK. So that is where a lot of the work uh is. In our case, we had uh about 800 subjects, about half and half males and females. We had seven days of observations and we had uh 1440 minutes and age between 31 and 96. Uh The data sets, the, the data set was, was, was pretty big. So what can we do in terms of analysis? Well, we can say you have an activity count for subject I at busy J I'm gonna try to use my fingers a little bit less uh subject IVJ time. T So this is say it in the middle of the day. I would like to know whether the activity for these individuals is related to age. And BM I, I'm asking a very simple question, is age affecting the level of activity in this case you could change this in, in your particular study you could be interested in. Suppose that I, I apply a treatment to an individual. Will they be more active or not? Right. That's a very, very good question to ask. And the good thing is that this is a silent observer that sits at the wrist. You don't have to ask anything. It just gets the data for you here. I just added t because it's gonna make for nice plots. But, but that's the que the question, the question is if you have the activity of individuals in the middle of the day, is that affected by age or by the BM I? Uh so I'm not gonna go with the details of how we do this statistically. But here are some of the results. So this is again midnight to midnight and this is the effect, this is midnight to midnight and this is the effect of increasing age and this is the effect of increasing BM I. So this is what's happening with an average profile of activity. So this is fifties, this is the average of for the fifties, 55 50 65 70. So what do you see what less, less activity in the morning? What else? Ages age is a huge effect, right? I'm not giving, I'm not giving you p values here because they are so small that you know, I cannot even write them down. So we are really talking about extremely small P values here just in general. This is, this is the, the good news here. A lot of the activity monitoring comes with very, very serious signal. What else do we see? Second, it looks like it looks like that. That's an interesting thing. We, we see this all the time, we actually saw it in monkeys as well, which is very interesting. Uh So it seems to be a kind of a primate type thing where there are, there is this deep in the middle and, and we see it over and over and over again. We're now we're producing many, many different uh applications. Uh So probably because, you know, in the morning, we have to do a lot of the work to get to work and things like that. And in the evening also, you know, this is a more, maybe a more quiet period where we are, people are more. Um the other thing that is very interesting is that as we age, we actually lose a lot more activity in the afternoon than we lose in the morning. So this is, this is something that has not been reported before. And it's something that uh it's quite obvious once you start to, to look at these things and we actually have data for even younger individuals. And it's kind of a sad thing to see that, but in the same time, it may, it may raise questions about when do you see the biggest effect of an intervention if you are to do an intervention, uh it could be that this may be the only place where you, you may get effect, but it could be that this is the place where you have the most leverage to have an effect. OK. So this is something that could raise all sorts of questions about research. What do you do? How do you intervene and, and, and what type of uh results do you get for increasing BM? I, we, we saw something similar which is that this is going from uh low BM I to high BM I. So we see that higher BM I corresponds to less activity, which is not extremely surprising, but we were able to quantify it. We, we were able to quantify that uh because these are comparable that about a 2.5 increase in BM I is about equivalent to five years of, of age. So just to, to get to an idea about, yes. Yeah. The, the scale, the scale is activity count is steps, steps. Yeah. But it's a very good question. It's something that it's deeply buried into the uh I, I can talk more about it. I I've been working very hard to try to get this into GS, which is more interpretable. But for now think about them as steps. OK. So these are some of the, some of the results that you can get with uh with this type of analysis. Very good question. You can, we haven't done it. I, I just wanted to do something extremely simple to kind of. But absolutely, you can do all sorts of things here. Interaction effects anything you want just as usual with regression, you can do that here because you have so much data as long as you have subjects is not a problem. You can, you can do interactions, you can do, you can do interaction between age and BM I you can do and I'll actually, I'll show you in a moment an interaction plot. Actually, we looked at interaction of time of day by age. So here I do the same thing. Don't, don't think, don't look too much at, at this. Forget about this part. I'm just interested in what is the average activity during the day? This is time t by age. OK. So this is the full interaction plot if you want and I'm gonna walk you through the full interaction plot. So this is, these are three plots here. This is uh this is women, this is men and this is the difference. OK? And again, this is midnight to midnight. OK. And this is 39 to 90. So I'm just gonna walk you slowly a little bit, you know, just to blue means low activity. Why? Because this is zero AM to about 7 a.m. Why? Because people are tend to sleep during that time. So, so people across the age groups are less active as they age. So as you go from bottom to the top here, people are aging, ok. So there there are deeper shades of blue here, but it's maybe hard to to observe red means activity. So going from red. So this is maybe 8 a.m. to maybe 6 p.m. You see how there is loss of color, right? It goes from red to yellow, that means less activity as people age. And you see that it goes faster to yellow to less activity in the afternoon. OK. So this is kind of a characteristic of, of the data again, we, we have seen this over and over again now in Haines and in many other studies. Um and we are now in the process of getting the UK bio data, Bio bank data, which has 100,000 subjects of activity so that we are in the process of, of doing that. Um We expect all of these things to kind of hold up. But uh we did hear uh women versus men because the first uh the first paper that appeared uh by and Haines said essentially the first result was women are more men are more active than women. That's their first result. And we just wanted to reproduce that. We, we wanted to, to see whether that's true or not. And uh our population is older than the enhanced population. So uh we cannot directly compare the two, but we, we can compare the two. So this is the difference between uh activity of men versus women. So this is the difference between these two plots and you know, don't try too much to, to understand. But blue means women are more active than men. OK. So what we obtain is that during the night, men tend to be more active. And even as they age, actually, there, there is an increase of activity of men during the night, especially in the morning hours. And Luigi said that this is probably men going to the bathroom. Uh II I have no evidence to the contrary. Um And overall, basically up to about age 60 65 it's a wash during the day. So it's basically about the same. However, after 65 women tend to be more active. So that that kind of goes exactly the contrary to the to the published results and seems to make sense. Uh also because women tend to live longer. So it could be that the, the biological age is actually younger in women than in men. But this is something that, you know, I don't know, I don't have evidence for that. I just have the data that seems to indicate something that's different from what uh is, has been published in the literature. And we have been able to reproduce this above 65 and above 70 consistently, this part. Uh I'm less confident about this part, about 65 and 70 we've been able to reproduce consistently over and over again. I added something a little bit for you because if you will work on these type of things, this is what you'll see. OK. So this is what the a software will give you to give you some plots and they kind of make a little bit of sense and they kind of don't make a little bit of sense, but I'm gonna walk a little bit through exactly what the data are. So this is a male age 73 BM I 36 and has 408 active minutes. So this is midnight to midnight again. And every bar here indicates when the device said that the person was active, 408 is the number of minutes. When the device says the sum of all these minutes that were deemed active by the device, this plot gives you the intensity of that activity, right? So this, this only tells you the person was active, this tells you how active. So it's one thing to say, I just walked for a minute versus I went and ran a mile. So the activity intensity of course, would be higher if you run versus walk. So these are, these are the the the different levels uh of the data. This is another example, a male age 72 I I chose somebody who is about the same age but much lower BMIBM I of 21. And this person had uh active minutes at 462. Pretty comparable with the other one. It's, it's interesting to see that the activity for this person is more compact. There are more periods that, of activities that are close to, to one another. I don't know what it means. I just see them and these are the types of questions that may be related to health may be related to the how active they are. But it's something that and imagine I could show you 800 of these plots. And if I look at the UK Bio Bank data, I could show you 100,000 of them, you get bored very quickly. OK, very quickly. Even with these two things, but it's useful, it is useful to kind of try to understand what is it that, that we measure and the same thing here for, for activity. So in this case, we are interested in modeling the fragmentation of activity we're interested in, in looking into how fragmented activity is and its fragmentation and intensity are they measurements are these measurements that could be used for characterizing health? We don't know right now, but we would like to, to get a better idea. And the way we think about it is that it's a process when somehow the person decides I'm gonna do nothing, I'm just gonna sit there, do nothing. But for some reason, there is a trigger or a need for the person to move, they may need to go to the bathroom, they may need to go watch TV or something. And then there is an activity that activity will have this level of intensity. Again, I'm not telling you what is the Y axis, there may be another trigger and the next minute and then the activity intensity would be higher. So these are the two components that we are trying to disentangle and see which one contains what type of, of information. So one of them is a simple question is the person active? Yes, no. The second one is if active, then how active. So these are two different, different things. And the reason for that is that a lot of the activity data and as some of you will, will collect this data, you will see a lot of the activity data has a lot of zeros because many times the person is just not moving. So you will see some of these ones from time to time and then they will be associated with a particular magnitude of, of, of that. And we went over that and we looked into some of the things about uh uh what you can do with this. So this is these are results for um activity uh whether the person is moving or not. So the intercept says during the night, the likelihood of moving is much lower than during the day, which is kind of obvious we were interested also if the person moved in the past hour and they have a particular intensity of movement in the past hour. What would that tell us about their likelihood of moving now? So, in other words, if you moved in the past hour, are you more or less likely to move now? And interestingly during the night, there is a positive association. If the person moves, they are more likely to move again during the day, there is no such thing. So that's an interesting finding about sleep and sleep behavior. Uh Probably this has to do with restless sleep or with the fact that the person is, is waking up, they're more likely to, to turn back again. And so this is, this is a very interesting uh finding for us at least. Um also age affects the, the likelihood of, of moving but more um in the second part of the day. So it reduces the, the likelihood of movement of moving in the second part of the day. And it seems to increase it though. It's not statistically significant in the, during the night. So that, that's something that we have to look more closely into whether older individuals uh move more during the night, uh than, than young individuals. Um And of course, we can do the same thing with BM I and with uh with uh other things. Um I'm not gonna go through all of this. Just wanted to give you a taste for just how much can be done. What kind of in depth questions we can ask about uh the various um types of activities. Um We, we tried to publish this paper in Jama a long time ago, it was rejected and there are now probably seven or eight New York Times articles saying the same thing. Uh We had 700 observations. The New York Times articles have about 1 to 2 observations, which is basically uh the the person who wrote the, that the article took some devices looked at their a Callie counter and compare it with what was given by the device. And um but what can you do? You, you do the best you can. So this is, this is, this is a device. So we did this in BLS A. So we looked at a calorie counter uh using an oxygen mask. Uh we had people at rest, this is their heart rate at rest and we had people then uh walking slowly and, and with increased. So this is basically the person on the treadmill where we very carefully looked into exactly uh the speed at which they, they walk and uh and this is their heart rate. So this is the heart rate on the x axis and this is the calorie consumption on the Y axis. And this is a very interesting plot. So you see here how the the resting state heart rate varies quite dramatically in this from about 45 50 to about 100. This is the reason why the calorie counters are not reliable. I was very curious why is it that they are not reliable? So the same reason is true here. If I take your, I don't take your, but if I ask you to take yourself your uh your heart rate at rest, it will be quite a bit of variability uh in, in even in this room. So let's see what happens. This is, this is one person, this is in black. We have these black dots here, this is one person and you see it's pretty linear, their calorie consumption with their heart rate, which makes sense, right? Your heart pumps more oxygen and just linearly pumps more oxygen. So that's, that's interesting. However, the line that is being used by devices is this population level line. This is the calibration line that is being used by the device. So the difference for this subject is about a calorie, half a calorie to a calorie per minute. That doesn't sound like a lot except when you multiply it with 440 minutes per day. And then you end up with a bias of about 700 to 1000 calories per day. So once we realized that we went very quickly to the published literature and we wanted to see what the public literature says and we did look and they report exactly the same differences half a calorie to one calorie per minute. So this is not unknown. This has been published and it's pretty well known what we did. We actually showed that, uh, we quantified specifically. So, uh, we know that in 43% of our population, the calorie Counter Counter, uh, was about, was off by at least 500 calories for 23% by about 750 for 9% by, uh, by 1000 calories. Um, anyway, I think, I think this is an interesting thing to, to, to know, I, I think it's, it's a little bit like one of the secrets that everybody knows, but it's good to kind of have there. Um, but don't stop exercising this, this is not what that is saying. Ok. So this is not, it's just saying that it's not counting your calories, but if you run for 10 more minutes than usual, it will still rank your activity correctly. Ok. So it's not to say that, uh, this is useless. It is to say that is not good for what it tries to do or what we try to. Uh, but I think it, it, it's a very, I use it all the time but more of an information about my level of activity, not about specifically how many calories because the problem is when people say, oh, I'm gonna eat 500 calories of this food and then I'm gonna go exercise for 500 Callies. That's when problems start to occur. Because that's not 500 calories you are eating and that's not 500 calories you're expending, that's when the real problems start to occur. Ok. Anyway, so, um that, that was my talk. Thank you very much if you have any questions, please. Um So maybe you guys have looked more into this. Uh I was just thinking about some of the data that we've looked at, for example, for hospitalized patients and their activity. Um This is a gross over, you know, simplification. But what we see, at least in the hospital is that there seems to be sort of kind of two populations of patients, sort of more active and less active. Um And here it seems like you're kind of classifying things by age and BM I and things like that. But if you sort of let the data kind of speak for itself, are there different sort of types of patients that kind of come out in the data? I was kind of curious about that, what we see we see. Uh I think it depends on how many patients you look at, right? So probably you can do a clustering to see whether that is the case and probably that that corresponds to the intuition of, you know, if you go to see somebody in the IC U, that person will never move, right? Um What we see in uh in populations in the wild is more of a gradation of activity from. So there is a lot of overlap, there is a lot of distribution uh going on. But I think it depending on the particular problem, you can, you can definitely let the data do some clustering for you. Personally, I still think that regression is, is, is a more refined approach. Uh because it, it will take into account a lot of the data in between and a lot of, a lot of the data that I look at is potato shaped. If, if I, if I plot eggs versus why it looks like a potato looks like a normal. So it's very hard to cluster that. Um That's because, you know, um our populations tend to be very heterogeneous in general. So there is a lot of overlap but if there is clear, if there is clear uh distinction between populations, absolutely just applies a clustering algorithm to that and, and, and go for it. I think it's very important to, to look at the next steps after that. What you now have two populations? How do you use that? And in that subpopulation that you have of people who are less active, say, what is it about them? Is it that they are, why, why are they less active? Are they all zeros or are they ones from time to time? And so I think that uh that could be something to, to look into as, as kind of a follow up step. Yeah, no sense of the extent to which there's variability in activity monitoring, done on the first on the ankle. So it seems like most of the devices that are partially available monitor and give a sense of whether that is because much there are big differences, there are big differences. Um People don't realize that, but there are big differences. Uh What I showed you here was at the chest and when we compare our average activity count with one at the hip is about 25% lower. The reason is that vibrations travel from your legs to your body and the head actually is the one that stays highest. So it gets fewest vibration. So as you go higher through the, you get fewer vibrations, that's not true for the hands because the hands just move by themselves. So they don't get vibrations from walking, they get more from just moving around, but they are different and they measure, they measure different things. There is a uh in the, a lot of the my collaborators, a lot of the people I work with and, and people I don't work with tend to, to move towards uh the risk because of a problem with compliance. So just imagine wearing something like that on the chest for seven days. It's not the mo the nicest thing. You know, people want to take a shower, they want to, it's not the nicest thing. And now, you know, the, the hand raise devices, they also have a sensor for non wear, which is, which is quite useful. So I think that a lot of people are moving towards the risk because of acceptability and because of, um, but that doesn't mean that it should be the only thing, you know, if you are interested, for example, in a wounded warrior who has a surgery on their leg, there is no reason to not have an el meter on their, on their ankle because you may be interested in exactly how their walking is changing over the course of the time. So I think I don't want to give one answer fits all because it really is more of a question of what is the question that you are trying to answer and, and that, you know, that that is the device that works best for, for that. Yeah. Also in terms of activity, one of the one to me because the times that I've seen people that think they just sitting down in one place. Yeah, you know, so that the accounts or whatever can go up and that's very different from somebody who's actually writing and you know, trying to get some kind of class for exercise. That's also different type of exercises and what the implication is for your, your health. And so related to that, I mean, someone can ask about rise and how you the extra balls. So you actually need, we actually had somebody who put the device on, on their dog we were able to discover that. But um it's true. Um Remember when the first thing I said, what, what the device measures does not measure activity. It, it measures the acceleration of the place where the device is actually placed. And that's something that it's very, it, it, we would like to say it's activity, but it's not, you know, the other thing that, that we know is that the device measures about, you know, if you drive, it measures like a third of walking, especially at the wrist, if it's at the wrist, then the vibrations that come through the wheel that that may, that is enough to, to give you about a third of the vibration. We realize that it, it's a very, so that's, that's so these are problems that you know, will come inherently with, with some of these devices. That's not to say that there's not tons of information in them. OK. So follow up to this. So what, what do you think is uh is this still moving forward in terms of uh the transaction between these technologies, different type of whatever devices and detecting different type of TV PS that I think, I think we are not there yet. I think, I think we're not there yet. In terms of, in terms of detecting, we tried very hard, you know, the first part, it's very hard to detect type of activity with these type of devices. Um I think that they are very useful in a supervised environment when, when you can actually see or, or, or videotape them. I, I'm actually moving quite a lot in the direction of videotaping. Uh I had somebody come to, to me asking me about the, there was a problem with some, some kids, you know, and they wanted to put a, on it. And I said, well, what is the problem? And they described to me what, what the problem was. And I said, well, why don't you just ask the, their parents to film them during their, with the, with the video camera? They never came back. So I learned my lesson. I learned my lesson next time I say no, we, we, we'll do some matters. But I think I think that the future is very bright for a lot of these things. But we have to be careful to not just talk about stuff, you know, if we, we, we shouldn't drink the kool aid, you know, we, we should, we should be very careful about what we measure and what we don't measure. And uh sometimes I think, you know, we, we are pretty comfortable with detecting walking, pretty comfortable who just left has done a lot of uh work on detecting walking, which is a big component of, of movement with some of the more intricate movements like my hands moving. Now. It, it's, it's very, very hard. Yeah, things about political feasibility. What we tend to find is that, you know, a lot of the devices, they're expensive or there's loss or managing infection control in the hospital um or different with extra personnel using different devices that might not have the reliability of other devices. And then all of a sudden doing a critical intervention and using that as a measurement. Do you have any suggestions on good devices? Things like cost? Yeah, I think the costs are coming down on, on a lot of these devices. So when, when I started one of the devices with like $1000 and now we are in the range of 200 to $250 for um I would say that the a graph GTX three is, you know, light is probably one of the, again, I have no interest whatsoever in that company. But it, it's a, it's a reliable device. The gene active device I showed here is, is also pretty reliable. All of them are reliable in the sense that they, what they have inside is basically the same thing, how they process the data and give it to you. That's different. So, so this number of steps and the counts. So you were asking about what it is, I don't know, I actually don't know because that is kept private from me. Of course, I do the standard deviation. I get exactly the same thing. So I know that what they actually measured is standard deviation. But I cannot say for Sure. So in the end, we can take the data, the raw data and, and transform it into things that, that are reliable, we understand what they are. So my recommendation is to stick to one device. And I, what I do especially uh if I try to generalize to a bigger population, I try to go to uh the basics of what it is that is measured and then express the measure in something that is translatable across. Uh for example, GS I, I would take this and I would say, what is the standard deviation over the minute that expressed in Jesus? I know that 0.5 G is about walking. 0.3 GS is probably slow walking and one G is probably uh fast walking for that person. So that to me is something that I would, I would do. But uh the research hasn't gone in that direction so far. It, it has been a lot more on. We don't care what the measurements are. We'll just do it because we can the relationship between things like wearable technologies and then reported patient reported outcome measures. It gives context. Yeah. But, but I think the other, the other thing that I would suggest that I think is very important is, is this idea and we have worked on several clinical trials, one of them was a clinical trial on heart surgery. And, and there is this question of how do you know, even if you don't have different devices. One thing that you can do is start with the device that you have and get data before the surgery and then keep on getting data relative to that. And then you can use something like a Z co relative to the baseline to see how their activity recovers over time. So that in that way, you kind of use the device as its own control and, and you do some Z scoring and hopefully that solves some of the problems. So that that's what I would do if I were to design a clinical trial from scratch. Um ok. Thank you very much. Folks want to stick around. We have a couple of residents presentation answers John, I'm sorry, I have to move. I had I hope you. No, no it isn't. Great. Also I will connect you who's uh who's next white sweater, long shirt, white shirt, black. Yeah. OK. Look her up around and, and some of the issues with them but actually Hi, I'm Leo. Ok. Are you presenting together? Ok. So I'm gonna actually mic her because she has like collars and stuff like that. Ok, cool. I mean I already know that you don't want to do this. I want to make it even worse for you. How often have you done this? How many times have you done it? Oh, really? It gets easier after the first. Let me tell you even if you don't think you're gonna do it again, I will never, I will not have a job that will require me to do this. You know, I'm button this is gonna, I wanna hide this as much as possible and you can button it back up. I know I'm sorry, I just wanted to have her. Is it on like in his live too? So everybody hears us. So what you could do it, you can mute it. And we're gonna go ahead, get, go ahead and get started. We're gonna be talking about evidence based recommendations for fall reduction in patients with Idiopathic Parkinson's disease. I'm Jacqueline Hall. This is Don Myers. Uh We're both um he's right for we have nothing to disclose. Um So the objective of today's talk is to describe the evaluation of a patient with Idiopathic Parkinson's disease, um who presents with kind of falls and also discuss some of the current research and recommendations for um falls reduction in those with IIC Parkinson's. Um Our patient is a 71 year old gentleman with cancer and hypertension and hyper Leia who came to the Falls Clinic for initial evaluation after being discharged from in patient rehab admission for a fall at home. Um He wants to know what else can be done to reduce his risk of future falls history, present illness. He had a an interior fall at home while trying to walk and talk on the phone. He felt like his feet were going to the floor fell forward and hit his head in the emergency department. The head was made for a review systems was negative except for fear falling and history of several falls in the past. Uh social history lives alone to level a home. Of course, three stairs to get in with just one rail while in invasion. We have, he was recommended to use a walker but before that, he was independent for emulation and a dia d without any device but continue to increase time. He denied any or was the drug use medication was uh physical exam were normal cradle nerves uh were normal except for mass like spaces and um hypophonia. His strength and rental status were both good and intact. Um tone testing. He did have some coal rigidity just in the right upper extremity, otherwise normal, um sensation was intact, perception, intact. Um Coordination did show some impairment in uh finger to nose and rapid ter emotion. Um Both were on the right of extremity but there was no, his reflexes were intact is unequivocal. Notably for transfers. He had to use both arms to get up. And his gate was observed without assisted device demonstrated a wide basis of support. He had increased hip flexion, poor initiation to start walking, reduce that length in both legs and also reduce arms, wing with flexion of the arms and wrist. He had decreased heel strike, decreased toe off and and walk turning some mild loss. Um So, uh differential diagnosis and some of their distinguishing features. Of course, this gentleman may have just had a mechanical fall and he checked on something. Um Parkinson's disease is high at my list because of this presentation. Essential tremor. Um The distinguishing features of that is that a tremor could be either action or postural. It's usually the predominant feature and these people don't respond to Parkinson's drugs, uh progressive supernuclear palsy, um distinguishing features is impair the vertical eye movement. Um They usually have a suitable affect and early gate instability issues. Um swing ponds as well. Um shy um drag or uh multiple system atrophy. Um They may have some autonomic issues. Uh cerebos are really common like a tax. They don't have a tremor typically. Uh again, they have this early date, uh complaints and swallowing. Coral vasal degeneration have more uh limb apraxia issues. Um They may have some cortical sensory abnormalities. Uh This tremor and early dementia is common in these patients. Also Louis body dementia have early dementia and some of them have more of the psychosis hallucinations and agitation that will usually present before before um presenting the motor symptoms. Alzheimer's disease. You can do dementia as a primary symptom, um drug induced parkinsonism exposure to dopamine blocking drugs. They typically don't have a arresting for asymmetry, uh target exposure to MP TP. Um Post in Supply Syndrome or vascular syndrome. Uh They used to have chronic hypertension like our guy did the wise progression and unilateral finding. Imaging can be used to distinguish the diagnosis. Um Some of their ideologies, medications wise, especially think about the antipsychotics, um especially the AIDS and Cols. Um lithium ate in Wilson, you see the tremor, right? So uh we referred this patient tour and we went to our clinic. He was diagnosed with uh the definition of that is a clinical syndrome caused by the progressive neuro degenerative movement disorder. Extinguished by the loss of neurons in the manger of the be know if the features are present without any other ideology, then it's considered a paic that's the most common over 80% theology patients. And what care staff um simply it's genetic protein folding or accumulation um and also uh mitochondrial dysfunction. Um This part, one gene on code for your body and that's associated with all of the, the age of 46 years old. Um And the difference between the body and Parkinson how quickly they're getting the bodies and levi disease. It's more diffuse early on versus for Parkinson's, it's really located the basal gala and then later on, they might develop it, more Laos treatments and symptoms should really raise red flags for you guys to think that something else and not was your response? And these are some of the genetic deficits or having most of them are pretty early on and um epidemiology. Um people on in the six decades, um prevalence of it's the time 1.5% huge 1.5 times more likely than women to be affected. The diagnosis. You are at two or three times higher, other was a gang. You have the generation of grand um this uh projects um you gonna cut it which is a association, an initiation um express mitochondrial dysfunction and are thought to contribute to these all we have. Um mhm Her which is the normal allegation seen on his most of these neurons in the substantial area has an imbalance between the dopamine and corner. Um For ray graphic features uh early on MRI you may see a decreased bit of the um or an a small hill sign is characteristic was the spot and so stay. Um this is seen on um the high res T two, right. Other options were in the pet tracers there going to have high but as people get more advance, more generalized win. Um so to make this clinical diagnosis, it is diagnosed by the motor signs, you need two out of three of rigidity tremor. Um Typically, no, because it's asymmetric, it'll be worse on the side that the symptoms first appeared on and then it may progress to have both sides. Patients may not remember. However, that one side was involved first. So don't count um The virginity is a velocity and direction since it gave to the unilateral what is called longevity versus your classmate, vision or spoil your group, she's just her arms playing drooling because they're not having spy swollen as often may also have four formation tests and they have some swing to you. They also have some pauses or arrest, which is character of the free. The asymmetric expressing tremor is the most common uh presenting to them. Um Usually in the hand, but it could be tongue, jaw foot anywhere and some people can report that they felt like they had a tremor before it was even visible. Um They actually, you know, we usually have the rolling forces hurt tremor, but they may have um the postural, it then hurts uh tremor or they may have both at the same time. It's not only is also known as alternate tremors because on emg you see these ultra burst in the agonous and antagonist muscle. Um This is typically worse when they get distressed or distracted, they're rushing. Um but it gets better when they're sleeping to do a task versus the postural um tremors uh which is similar to essential tremors where it's actually arresting. Um because the central travelers can also be in a patient with Parkinson's hand writing uh is a good way to kind of distinguish them. Um This big uh a patient um make a so Parkinson versus someone with just as essential that character six small other uh physical exam bindings are the soft voice or mumbled and speech I'll talk with. Uh they used to have uh small ratings we talked about and um G balance and partial instability issues uh has to violate research shows it may be more common in later stages. But I think that's still up for debate because patients are still found, um, to benefit from um, interventions actually in the earlier stages rather. Um, it's um interesting though that when progressed, their gate problems in the may not respond to medications as well. It's a characteristic of shuffling or president walking speed, the short stuff, shuffling step, as we all know in the semi and posture with the flexed elbows, flexed wrists. Um, but um, as we know, they have difficulty of initiation, things are getting from the chair taking that first step, they may have decreased spatial to the and walked her in, got you this. Um And the way moving movement initiation is common, especially with multitasking distractions if they're rushing or they're anxious and this is called the freezing uh where they lose, um, push off as this image here described, um, they're not able to get the push off and take the next step. Um, as our gentleman described, he felt like his feet or stuck to the floor. Um This is associated for many patients of um causing falls and also being associated with the fear of falling. Mhm. It may occur in what's called off periods which is in between. Um, medication does kind of the trough of the medication concentration. Um So consider uh asking neurology or um yourself to talk about, um, adjusting their medications and see if that helps. But for some people to still purchase, even, even though you've made adjustments to their medications and they may later stage. Um, so triggers for freezing, uh, because we worry about falls and you want to, and remind them to be extra careful when they're doing some injection changes. Um, things like doorways, uh, thresholds, obstacles, uh, two and four patterns, stress crowds, these things can cause freezing to be. And so if patients learn how to prevent them from having or how to manage it and kind of find some composition strategies, they may be less likely to fall. Um So the four essence of acronym uh you can teach patients um to stop, don't force your way through the freeze because if you do, you're probably gonna have that forward fall, um stand tall um doing a side to side sway and also step long, we'll talk more about that as these big steps. Um rather than the short shuffling, that's a big focus on that. Um So to the thing, because this is a serious problem for multitask in it, what you know, you can trigger that and exam. So you actually ask them to work and then you as a patient, you ask them a question, there's nothing to do with it. So you are kind of problem to multitask and then you can trigger what this this method. Yeah. And, and then just teaching them you know, to avoid multitasking if they have fear of falling or before, you know, um, because I, there are a lot of people who report a lot of anxiety issues related to this issue specifically. Um, so the Strothers Ferguson Center in 2005 did some, um, research on s and injuries in pretty good size population of over 1000 most people have her's diagnosis for at least seven years. Um, 55.9% have had at least one fall in the past two years and they had pretty high morbidity and mortality. Um The uh 65% of time injuries are fracture in 75 health care services. So as you can imagine, there are some high costs, um age rage changes of course, can contribute to increased false risk, but just having the diagnosis of Parkinson's associated with um two times increased risk, of course, they have these extra risk factors um following with posture, instability, freezing, et cetera. Si course the majority of people fall forward at 45% direction and I'm sure and a lot of this work, some people will report that the wearing off effect of their medications are associated with falls. But you know, they also, um and this can occur when you walk in trade asking, you know, reaching for any um data that you've seen in terms of the directionality of falls and risk of injury. Oh, in terms of like what type of injuries are associated with it better price than this. Um, you know, I didn't see that recorded but I'm sure, I'm sure there's, it seems like fun backwards. Yeah, I would. Yeah. So, and, and one of the important things about this patient population is that not only they have problem that been involved so, prisons, not that they have a decreased post reaction so we can't catch them so they cannot catch. So. Yeah. Um and so some of the um therapies will talk about us to specifically work on that issue. Um So even though this is a um diagnosis that's based on their motor symptoms, we cannot ignore the fact that people also have some um pretty significant that may come first actually. And um they can really affect the quality of life. And so we try to address these issues over 50 per cent what whether it's some isolated minor issues to severe dementia. So, you know, getting your like involved. Um some people have psychosis and it's often related to medication. Of course, as we get older, depression and apathy is pretty common. These patients are um they may have had it preexisting but then also now dealing with a new diagnosis that affects other. And as I kind of all your eyes face, some anxiety, social panic attacks are associated with under medication. Um But actually the disinhibition we treat um other normal symptoms, constipation or like hypertension and I kind of run through time, um, sleep disorder. Um, some people have, is called the so, could be particularly disturbing to family members. Um, because, uh, um, also interesting in fact that came across with that two or high risk of invasive melanoma and Parkinson's, it didn't seem to be related to medication. Um, um, the olfactory dysfunction is actually, um, been found to be one of the old, um, markers for, um, because the, um, the loss and it also affects the olfactory system. And so 70 to 90% of our patients are found to have decreased sense of smell. Yeah, they um they come across with this upset smell test, uh which is uh commercially available. Um I think it's like three or four bucks that patients can do themselves in the waiting room. It's basically a scratch and, and you do very sense. It looks like this here. And interestingly, some people have found that there are certain smells are particularly um homework for Parkinson's. I still other things to think of these people because of course, this may affect their appetite, taste. And uh so as I talk about the speech loss, it's called hyper kinetic art. Um originally, data suggested that this occurred um mid or late stage and that people have speech influence problems. Um but it seems like it's actually occurred from paper. Is this reduced volume monotone and hoarseness that comes first before they have issues with their speech inflections and then also a tremor and articulation with speech was doing music in this short rush of patients. And also I found that they for John Eng studies have found that the Agnes and anti muscles are we are also and of course, how of speech can affect their jobs or interpersonal communication? It can be very frustrating for it in the play, especially with kind of anxiety, embarrassment withdrawal. So uh speech involvement is important for these patients. Um For functional assessment, there's a lot of things you can look at uh manual dexterity. Um The time get up and go test really easy to do in the clinic. Um You ask them to get up from a chair with armrest and go 3 m and you time them five minutes at the stand, you walk past balance. Um The PDRS pull test for cultural stability um real easy. You just have a patient stay with their feet, slightly apart, eyes open, you know, they are prepared for um for getting shoves. Basically, you tell them you're gonna do that and tell them you want them to catch yourself, you go stand behind them and you give them a nice shove, obviously be prepared to spot them. Um And you can grade that and that's associated with um with high risk for falls in the community. So this is something very, very easy to on your own. Another one, the push release test, uh which is a component of the um best uh test um very similar. Um You have the patients stand with their few arms apart and you give them a push and also some scaling. Um data has suggested people have an abnormal um push release test should be candidates for intensive and multi factor intensive programs including optimizing their medication specific education. So those are things we can disease progression. Um Usually the symptoms begin insidiously uh as I mentioned, unilaterally, but become uh bilaterally with time. Um And in later stages, it's when you start to see more of the phase changes to the um the disease duration from diagnosis is 15 years. So it's a long time, you know, still some of our. So we're going to discuss some of the rehab specific management. We intentionally kind of left out some of the medication discussion and just try to focus more on things either up or coming in terms of we would be managing more of. So with early disease, usually you want to address balance, training, lower limb strengthening. Some different activities could include dance, tai chi and video gaming. All of which like the video gaming, we have been demonstrated to improve balance as they encourage the wide base of support. Stretching programs are also important educating the patient to perform consistent aerobic exercise and then resistive exercise have been shown to some of the Parkinson's disease, motor progression on the unified Parkinson's disease rating scale group exercises such as community classes have been shown to increase quality of life and neurological conditions, especially Parkinson's with increased continuation of exercise. After the program completion, when the disease continues to progress to the sub acute stages, there's more axial rigidity. This should be addressed with flexibility. Training. Always continued to encourage cardiovascular conditioning, continued strengthening. They may start requiring some training with external cues. Any cognitive movement strategies, body weight supported treadmill training can assist possibly in maintaining activity as they become more chronic, they're going to need more and more assistance may progress them to will walkers, brakes, stabilizing base more and more cues requiring visual and auditory queuing strategies such as portable metronomic devices. These can assist with gate initiation and gate speed. So then we'll talk some of the other treatments, virtual reality training, rhythmic auditory stimulation force exercise, virtual reality training is using any visual cues that can improve. Git one of the studies we looked at looked at treadmill training plus virtual reality training. It showed that there was reduced fall rates when compared with treadmill training alone. Their virtual reality system was a computer generated simulation projected to a large screen which was specifically designed to decrease the fall risk in older adults by having them practice real life challenges, obstacles pathways and adding distractors. The gate aid is a more portable training, virtual reality training device. It comes with glasses. There's a picture on the next slide that it projects this high contrast checkerboard pattern up in that other picture there onto the floor. So that way improves the patient's walking ability. Here's an example of the gate aid. So it's definitely a more portable one that can attach to their belts with the glasses. Rhythmic auditory simulation was initially developed by thought mcintosh and rice at Colorado state. It utilizes the physiological effects of rhythm on the motor system to increase the efficiency of controlled movement patterns. During rehab with Parkinson's disease, they use that employs music or metronome based cues to improve the game performance and modulate variability. This enhancement is mediated by an entrainment effect whereby the movement frequencies and motor programs and train through this rhythm and anticipatory queuing. It has been demonstrated to statistically significant improvements overall, excuse me, observed in the mean gate velocity cadence and stride length problem with these things is that although they are helpful when you are a user, he was. Mhm They don't not generalize that the moment you remove the devices. Mhm. So there there's so much carryover. But um here here uh Alex is from listen heavily doing the music. So basically playing, playing with rhythms, civilization of rhythms and also playing with drums. Kind of very simple. Yeah. Mhm But that's the, that's the main problem though they have there's no matter. Mhm. So in some of the research we read about the Ely Me device, there's also a smartphone application that the patients can utilize. It allows them to pick the different auditory queuing rate. They are very slow, slow, fast or very fast. This requires the use of a Bluetooth for the patient as well or they can have the wireless glasses, wireless headphones. This is a video of the patient utilizing. I don't know if it'll work. If not, I'll skip forward, we need to get forward anyways. Where are we at time? Forced exercise is another form of treatment. It's a mode of aerobic exercise in which the exercise rate is augmented mechanically to assist the participant in achieving and maintaining an exercise rate that's greater than their preferred voluntary rate of exercise. The participant should be actively contributing and not just being moved passively through the motion. Originally, this was noted with tandem bicycling, they found that it could improve motor function and bimanual dexterity. The data seems to indicate that this forced exercise leads to a shift in the motor control strategy from a feedback loop to a feed forward process type of exercise equipment. Example that the patients can use for home or use therapy would be a the cycle and it's able to replicate the 80 to 90 RP MS that have been shown in studies to improve the motor function. This is an example of the cycle. The Lee Silverman voice treatment, big and loud is another treatment. Traditionally, therapy for hypokinetic dysarthria focuses on rate articulation and pros with only modest and short term results, the tendency to overestimate the size and strength want to pause that. Sorry. Oh Let me go. Sorry guys. So I say if they are interested on learning more like some of your research. Oh yeah. And no 3000 they work right side. No, I didn't fuck. OK. So Ray and his colleagues created the LSBT treatment. It's an exclusive target of increasing the amplitude, including loudness of the speech motor system and bigger movements in the limb motor system. The focus is on sensory recalibration to help patients recognize that movements need to have increased amplitude. They are within normal limits still, even if they feel too loud or too big for the patient. The goal is to teach participants to carry over and sustain the bigger movements and louder speech in their day to day activities. This targets areas in the basal ganglia through repetitive activation of the motor regions involved in the movement amplitude, consistent with principles that promote activity dependent neuroplasticity. The loud therapy improves vocal flow, vocal fold, abduction for optimum loudness and quality. Without the undue strain benefits include improvements in facial expression, improvements in hoarseness, speech, intelligibility, swallowing and hypophonia generalized effects have been shown to last up to two years. Individuals with mild to moderate disease have the most positive treatment outcomes. However, patients with co occurring, mild to moderate depression and dementia have succeeded in treatment as well. LSVT big features multiple repetitions, self queuing of large amplitude movements such as arm swing or a step to improve movement initiation. There are sustained multidirectional and functional movements while seated are standing early. Parkinson's patients have been found to have improved quality of life on the PDQ 39 scale. Improved date speed, bird balance assessment, motor symptoms rating. They have been shown to have better improvement on the UDS motor scores as well as a tug and 10 minute walk. In a small study of the LSVT, there was not shown to cause lasting improvement greater than four weeks in the motor scores during the off phases or decreased amount of daily off time in the advanced Parkinson's disease. But it may improve the motor function while on the liva dopa. This is the last one that was just a novel therapy. I hadn't heard of yet, but it was using MRI guided focused ultrasound trans transcranial thermal. So under MRI guidance, they direct a focused beam of acoustic energy to coagulate small areas of the brain. It's currently being studied to treat the different symptoms lesion in specific areas with the tremor a lesion in the thalamus, with the dyskinesia lesion in the globus poly or sub nucleus eks lesion in the pathic tract. Currently, it's only being utilized to treat one side of the brain though. So just our summary slide for the patient, we would recommend regular exercise that has been shown to help the brain produce productive growth factors and decrease early Domenic neuron death exercise should be adjunct to medications to provide even more benefits. The patients. We should encourage our patients to exercise and provide therapy referrals often and as proactive rather than reactive, encourage a patient to be creative. Lots of fun options out there for our patients. Yoga, tango, boxing. You name it so sorry. I ran over guys. Ok. A bit really has a machine and then he uses it. Took over for the.