Stephanie Cernera, PhD 

Doris Wang, MD, PhD

Stephanie Cernera / Doris Wang

For our August session, on August 31st at noon ET / 6PM CET, we hosted Stephanie Cernera and Doris Wang.

Stephanie Cernera, PhD, is a postdoctoral research fellow in the Starr lab at University of California San Francisco. She told us about “Personalized chronic adaptive deep brain stimulation outperforms conventional stimulation in Parkinson’s Disease”.

Doris Wang, MD, PhD, is a neurosurgeon and (recently promoted!) associate professor at the University of California San Francisco. Apart from her scientific presentation about “Adaptive Deep Brain Stimulation to Treat Gait Disorders in Parkinson’s Disease”, she gave us a glimpse at the “Person behind the science”.

Please follow us on Twitter for updates about future sessions!

Watch the recording with our interactive transcripts below!

and Doris Wang. And I would also like to thank my co-organizers, Barbara Hollander and Gerard Smeyer, and also to Andy, who is very supportive of this talk series, but unfortunately, today, he couldn’t join live. So today’s session will start with Stephanie’s talk on personalized chronic adaptive deep brain stimulation outperforms conventional stimulation in Parkinson’s disease. Stephanie is a postdoctoral fellow in the STAR Lab at the University of California, San Francisco, and she’s focusing on adaptive deep brain stimulation research. And then this talk will be followed by Doris Wang, who will speak about adaptive deep brain stimulation to treat gait disorders in Parkinson’s disease. And Doris Wang is an associate professor and a neurosurgeon also at the UCSF. And she specializes in surgery for patients with movement disorders. And besides her more scientific talk, she will also give us a glimpse at the person behind the science. And just to repeat a few rules for our attendees, everyone is muted, but you can all ask questions after both talks are finished, because there is a Q&A session at the end. But you can also type your questions in the Q&A channel during the whole event so that at the end, we can read them for you. And other attendees can also vote for them. So basically, those questions that get the most votes will be preferred to be read out because of the time constraints we have. But of course, if you raise your hand, at the end, we will unmute you and invite you to the stage. So I think I covered everything. So we would like to start with our first talk. And Stephanie, please welcome to the stage. Thanks. And you guys can see everything. Fabulous.

Well, thank you everyone so much for being here and thank you for the invitation. I’m really excited to present our recent work in adaptive stimulation in Parkinson’s disease. But before I get to the results, I don’t know. Oh, perfect. I do want to point out that this was a massive team effort. Dr. Karina Earn, myself and Dr. Lauren Hammer all share co-first authorship of this paper and doctors Simon Little and Phillip Starr are also co-senior authors and we received in a massive amount of help from the other individuals listed on this slide. I just get to be the face of the team today, but it was a huge effort. all really excited to get these results out. A brief background of adaptive stimulation in Parkinson’s disease, patients who undergo DBS with PD typically still take levodopa medication, albeit to a lower amount than pre-surgery. And as they transition throughout their medication cycle, this presents different situations in which they may need different stimulation amplitudes to optimally control symptoms. For instance, they may feel off or have slowness of movement or bradykinetic and to suppress this they may want to increase stimulation amplitude. Conversely, they may feel on but have unwanted movements or dyskinesias and to avoid this they may want to decrease stimulation amplitude. But with adaptive stimulation and in our case we’re using neural biomarkers as control signals, we can detect a motor fluctuation with a neural signal and appropriately respond. So for instance, in this case, this person’s feeling bradykinetic while cutting their peppers at home, and we can detect a change in the neural signal. In this case, it’s dropping below a threshold, and we can increase stimulation amplitude to improve movement speed. So with adaptive, we can automatically adjust stimulation amplitude and response to motor fluctuations, improve symptoms like bradykinesia, potentially avoid stimulation-induced adverse side effects like dyskinesia or dysarthria, which is speech impairment. And lastly, but most importantly, with adaptive, we have the potential to increase the quality of life with patients in Parkinson’s disease. And there’s been two control signals used in previous adaptive studies. The most common has been subthalamic beta activity. And this is an anti-kinetic frequency band. It’s associated with bradykinesia and rigidity. And in this example, this participant is off medication. You can see this peak in the beta power band. And once they’re on medication, there’s a suppression of beta.

Finely tuned gamma has also been used as a control signal. It’s a pro-kinetic frequency band. It’s associated with the on medication state. And we’ve previously reported observing finely tuned gamma in the motor cortex of patients with Parkinson’s disease. In this instance, when the patient’s experiencing dyskinesia, you see a peak in the finely tuned gamma frequency band that is suppressed when there’s no dyskinesias present. And although previous studies have shown that adaptive stimulation may be more efficacious than continuous DBS, they have been either short perioperative or in clinic studies, and we really aim to expand adaptive stimulation to naturalistic settings. But first, we did want to identify potential control signals across the motor network, So both STN and sensory motor cortex, we didn’t want to limit ourselves to one recording site and across the entire frequency range. So again, not limiting ourself to either beta or finely tuned gamma. We did want to identify control signals during active stimulation since our patients toggled between lower and higher stimulation amplitude for optimal symptom control. So that would be the most realistic in naturalistic settings. And finally, we did want to perform adaptive stimulation in the home environment in patients with PD. In our case, we are using the Medtronic Summit RC+S device. This is a bidirectional device, meaning we can sense neural signals while simultaneously delivering therapeutic stimulation. Patients were implanted with the DBS lead in the subthalamic nucleus, as well as chronic electrocorticography strips of our sensory motor cortex. So we can sense from both these areas, And additionally, the RC+S can perform embedded adaptive stimulation. There are three patients in the study and before entering the adaptive phase, they were clinically optimized on continuous DBS on average around 22 months before starting adaptive. And with them, we identified their most bothersome motor symptom on continuous, which for two patients was bradykinesia, that slowness of movement, and for one patient, off-medication dystonia. specifically in the lower limbs. And additionally, we identified the most bothersome body side, which was unilateral for one patient and bilateral for two patients. So we did perform adaptive stimulation in five hemispheres. And with patients, we also identified the opposite motor symptom. So this was the motor symptom that really limited the therapeutic window of continuous TBS.

We assess symptoms using a variety of measures. We use an objective wrist worn wearable to assess bradykinesia and dyskinesia. We use subjective measures such as the motor diary, which asks patients if they’re having a symptom every 30 minutes. And most importantly, we developed a daily questionnaire in our lab. And this probed patients about the amount of daily time spent with the most bothersome and opposite motor symptom and other common PD symptoms, because we didn’t want to make those worse with adaptive stimulation. And we also used a validated daily quality of life questionnaire. The overall goal of our pipeline was to customize the algorithm to an individual patient’s residual symptoms. And we hypothesized that this would lead to individualized biomarkers or control signals. So our first step was to characterize those residual symptoms. And next, we titrated to a higher and lower stimulation amplitude in the off and on medication state respectively that led to optimal symptom control. Next, we performed neural recordings for biomarker identification, first in the clinic at these different medication and stimulation levels. And these were standardized motor tasks that were paired with a wearable monitor. Next, we perform at-home recordings, one day at the high and one day at the low stimulation amplitude. And these were paired with individualized motor diaries as well as the objective wearable. And again, all performed during active stimulation. And finally, we optimized the adaptive algorithm once we knew the control signal. And finally, we tested adaptive DBS versus clinically optimized continuous DBS in naturalistic settings, so in the patient’s home environment. And this was blinded, randomized testing, so patients did not know if they were on adaptive or continuous. And they would switch themselves every two to four days using a graphical user interface that we developed in the lab. And we did this until about 30 days were met in each condition. Just for time’s sake, I’m only going to show the data driven biomarkers from the at-home recordings. And since the RC+S uses a linear detector, we aimed to find the frequency bands that best distinguish between the presence and absence of the most bothersome symptom at home. And to do this, we used an LDA, or a linear discriminant analysis. And what’s plotted in these color maps is the area under the curve. And a higher area under the curve means the more predictable a frequency band is. And what is plotted is the color map of each recording site, so the STN and the two cortical montages. And the y-axis is the patient’s hemispheres and the x-axis is each of these frequency bands. And we found that in all three patients across the five hemispheres, that finely tuned gamma at half stimulation frequency best predicted patients’ most bothersome symptom in the home environment. And the arrows are just denoting the eventual control signal used in our adaptive paradigms. And you can see that for two patients, a cortical montage was used or cortical finely tuned gamma and for patient 2 the sub-dynamic finally tuned gamma was used as the control signal. But you might be asking yourself well what is finally tuned gamma at 65 hertz? Well it’s becoming increasingly more known that patients have levodopa induced oscillations in the motor system and this is just an example from patient 1 in our study as we were toggling stimulation amplitude over a few minutes. And you can see that when stimulation is at relatively low levels, you can observe levodopa-induced finely-tuned gamma. So this patient is on medication and they’re on state, and it’s around 80 hertz. And then as we increase stimulation amplitude to around 2 milliamps, this levodopa-induced finely-tuned gamma becomes entrained of 265 hertz or half stimulation frequency. And we also report that finely tuned gamma fluctuates with the most bothersome symptom during stimulation at a constant DBS amplitude. So this is a recording over around 10 hours at a constant DBS amplitude, so nothing is changing and again in patient one. And when the patient marked the intake of levodopa denoted by these red dashed lines, around 45 minutes to an hour later, you can observe an increase in finely tuned gamma, again, denoting that they’re in their on state. And then when the patient marked the onset of their most bothersome symptom, denoted by these black lines, finely tuned gamma is now suppressed. Since subthalamic beta has been used in previous adaptive stimulation studies, we compared the predictability of beta power to finely tuned gamma. And we found that finely tuned gamma across patients and hemispheres demonstrated a higher area under the curve compared to STN beta. So it distinguished best between the presence and absence of that most bothersome symptom. And when we added STN beta to finely tuned gamma in our LDA, we found that it did not add additional predictive value to finely tuned gamma across patients and hemispheres.

An example of our adaptive algorithm again over an entire day in patient one. And as a reminder, their most bothersome motor symptom is dyskinesia and their opposite symptom was, their most bothersome motor symptom was bradykinesia and the opposite was dyskinesia. And again, the most discriminative biomarker was finally tuned gamma at 65 Hertz in the motor cortex. And that’s what’s plotted in this first subplot here. And again, when the patient intakes levodopa denoted by the red dashed lines, About 45 minutes to an hour later, we see this increase in our control signal or finely tuned gamma denoting the on medication state. And to avoid any dyskinesias, we decreased stimulation amplitude to a lower level. And once our biomarker went under our lower threshold, denoting the off medication state or hypokinetic state, we increased stimulation amplitude to suppress or improve any bradykinesia. What we found is that across all three patients, adaptive stimulation shown here in the orange bar graph significantly decreased the time spent with the most bothersome motor symptom in the home environment. And continuous stimulation is shown here in blue and you can see across all patients adaptive significantly decreased that time spent. And this was across around 30 days in each condition. And again, in naturalistic settings. So super exciting to see this. And we decreased the most bothersome motor symptom, not at the expense of the opposite symptom. So for patients one and two, there was no difference in the opposite symptom between the two paradigms. And in patient three, adaptive actually decreased the time spent with the opposite symptom as well. Additionally, in two out of three of the patients, we found that adaptive BBS did significantly improve quality of life. And just to quickly look at the algorithm dynamics, since previous adaptive studies used really quick algorithms that led to a decreased total amount of electrical energy, which authors hypothesized would lead to less stimulation-induced adverse side effects. But in our case, we were tracking motor fluctuations, so we were playing a slower game. And what we found is that patients spent about 75% of their awake time at a higher stimulation amplitude compared to around 25% at the lower stimulation amplitude. And as you’d expect, this led to an increased total amount of electrical energy delivered. However, again, this was not at the expense of the opposite symptom. And this is because we were really delivering targeted therapy when patients needed more stimulation.

conclusions in future directions, adaptive stimulation reduced the most bothersome residual motor fluctuations in naturalistic settings. Finely tuned gamma in the SDN or cortex were reliable biomarkers and this was facilitated by multi-site sensing. For most participants we did deliver more electrical energy but again this wasn’t at the expense of the opposite symptom. And our next steps we are performing a two-month clinical trial and this will be double blind, randomized, and instead of those quick switches between the two paradigms, it’ll be one week, well, four weeks per condition. And this is a really brief overview of the study, we sped right through it. So please check out our preprint on medRxiv, reach out to us, ask questions. We’re happy to talk to everyone and receive feedback. And I just want to quickly acknowledge the participants in our sensing-enabled devices at UCSF. They’ve all been amazing, and the STAR Laboratory, my funding sources. And if you have any questions or comments, please feel free to reach out. Thanks. Super amazing. Such a cool project and also great presentation, obviously. Thank you so much. So maybe just as a reminder for our participants, we will not take questions now, but just keep them in mind for the shared discussion later or type them already into the Q&A chat below. And we’re quite excited now to hear from our second speaker, Doris Wang. All right, thank you so much for inviting me and a fantastic talk from Stephanie and it’s a great introduction to actually what I will be sharing with you all. So going to share my slides. All right, everybody see this okay? Wonderful. Okay, so, um, you know, I was asked actually for the first time as kind of the senior speaker for our talk series, which makes me feel old but I guess I have been in this field for now five years. And again, the biggest disclosure I want to give you all is that there’s no correct path, and everyone, you know, gets where they are following their own path, and it may be tortuous, it may be really challenging. But I just want to share with you all, you know, how I got to where I am. So, one of the first topics I want to talk about is managing failure, because you know whenever you go to a talk, you know, the speaker, you know, flashes all their funding sources, all the wonderful results that they have, but behind every, you know, results and every success, there’s probably, you know, 10 times as much failure that the patient had to endure to get there. So I think one of the things that was really challenging and probably continues to be for young faculty and people going into science is, you know, the grant cycle, paper cycle, getting rejections. You know, we all probably grew up being pretty much the top student, you know what you need to do to get good grades, and it’s a very defined path. But in graduate school, in science, in and medicine and success as a scientist, you know, it’s a lot of things are beyond your control. So just give you an idea. So when I started my fellowship in functional neurosurgery here at UCSF from summer of 2017 to summer of 2018, I applied for a lot of grants. I knew I wanted to be in academia. I wanted to do type of research that Stephanie’s mentor, Phil, who was also my mentor, started. And I applied for K23, I applied for the Burroughs Wellcome Foundation grant and a bunch of other grants. Basically, I didn’t get any of them. And then, but I was at the end of the fellowship, got a small, very small foundation grant, but that’s really important. Just that small positive feedback, you know, can give you enough money to hire somebody when I started my lab. So when I started as faculty at UCSF, you know, I replied for the K23, got not discussed again. And then you can see what I did during my first year as faculty. I applied and wrote incessantly. And a lot of these were not successful and I would get quite far in the application process, but it’s very disappointing not getting them. But I guess one thing that helps me to keep going is that I knew the science was interesting. it’s important, something I want to keep going. And plus it’s a learning process. You know, you get these grant workshops and grant writing pointers from others, but it’s just like going to school. You know, the more you practice, the better you get. The better you get at communicating science, picking out like the key facts and how to write science. So luckily that year, later that year, I was able to get a K-12 grant from the Neurosurgery Foundation and then the NIH, and then got the BWF, several UCSF Catalysts and well neuroscience grants. And this really set me up to grow my lab and pursue the projects that you’ll hear about. And this cycle continues. So even in the past three years, I lost count of how many grants I’ve applied for and have failed to get. And one advice, my mentor, Phil Starr, has always given me is when you get your score sheet or bad news, rejection and whatever, allow yourself 24 hours to wallow in self-pity or be sad, be angry, whatever you need, but then just kind of move on from it. Just know that none of this is personal And anybody who’s ever successful had to deal with all this before you, and you just keep going. And in science, especially, I think persistence and perseverance and resilience are probably the highest qualities that can predict success in the future. And the other thing is, you know, being passionate about what you’re doing and finding what you’re doing really, really interesting. So there’s a lot of failures, and I’m sure there’s a lot of failures I will encounter, you know, in the future as well. But just know that this doesn’t define you. This is not, you know, your whole sole purpose in life. So I find that kind of advice useful. And then the other are some things, you know, there’s so many topics that I can give and get into. One thing, you know, since I am a neurosurgeon physician scientist, I think, you know, having a clinical practice actually, while it does compete with my research interests somewhat, it is also complimentary in a lot of aspects. You know, when science isn’t going well, and science kind of what I described with people is that, you know, it gives you, you know, satisfaction and really wonderful results if you get them that can last like months and years. And it’s long-term happiness, but the times that you get them are very small. But in terms of clinical neurosurgery, what I do, you know, in the OR, it’s instantaneous gratification. So that’s actually a really nice compliment when I’m not getting the results I am for my scientific work. So the reason I was drawn into neurosurgery since basically high school is, you know, the ability to study the human brain and unravel what this magnificent machine is doing, understand it and understand how disease and injuries can prevent it from functioning properly and how we can actually as physicians try to restore this complex circuit. And while my main practice is in stereotactic and functional neurosurgery, so that’s using ablative or neuromodular techniques to treat movement disorders. Some of my practice also involves other types of surgeries. Like I do quite a bit of spine surgery, general neurosurgery, sometimes vascular tumor. And it definitely keeps it really, really interesting. And that’s one of the reasons I love neurosurgery. I love being in OR, I love operating. And the variety of cases that I get to do is quite amazing. And then the other thing that’s really drawn to me about neurosurgery is all the cool toys we get to play with. We have, you know, basically we’re at the forefront of, you know, technological innovation. you’ll see some of the things that we’re doing in the laboratory from the research perspective, but even from a surgical perspective, using robotic surgery, advanced spine instrumentation tools, navigation. So it definitely keeps the day interesting. And then I think the other thing is that neurosurgery affords an amazing research opportunity. when I’m operating, doing a deep brain stimulation surgery, I can study the subjects and the patients I’m treating, you know, looking at doing intraoperative tasks, for instance, and some of the work I do is very similar to Stephanie’s, you’ll see, and even after operating on them. Not only are you treating their symptoms and diseases, actually gives an amazing opportunity to study how the brain is functioning in the natural environment. So, you know, with the professional side, it looks, it’s pretty busy. And then, you know, everybody asks, how do you balance work and life? You never balance it. It’s just, you find time for both. And, you know, at different points of your life or different hours, even within the day, you prioritize one versus another. And it’s really hard to have it all. I think to kind of keep going and also to find fulfillment in both life and work, you have to be surrounded by the right people in the right environment. So, you know, there are many hats that I wear on a daily basis and my schedule varies and it’s pretty crazy day by day. So, but you know, the core things that make me happy is what I do for work, obviously, in terms of doing operations on patients, running a research lab, teaching the next generation of physician scientists or scientists, treating patients. That’s also like one of the most rewarding aspects of my life. And also, making cool discoveries, understanding human neuroscience. And then outside of it, you definitely have to have passion in life And this is like, you know, what I used to do before I had a kid, obviously, and before life got crazy. But, you know, I think traveling, finding exciting activities right now, do like surfing, diving, yusu rock climb, stop that a while ago. But, you know, finding something that’s fun and take your mind away from, you know, the everyday work life or everyday life. it’s really, really important. And then, you know, one thing that cannot be underscored enough is finding good mentors and advocates for yourself. So throughout my career, I had an amazing group of mentors who, and advocates who, you know, help support me when things get tough, you know, listening to their advice, both in life as well as in the career, and who can advocate for you, who can promote you, can point you in the right direction. And they’re also amazing role models. So, this is my undergraduate research mentor, Angelique Bourdais, Arnold Krigstein, my PhD advisor, Phil Starr, my current department chair, Eddie Chang, who is a true inspiration for what one can do as a neurosurgeon scientist. My husband, who’s also a neurosurgeon, Aaron, And then, you know, Paul, who moved away, Jill from our Movement Disorders Center, my former chairman. So all these people are really important for each step of my career in, you know, molding and shaping what my practice, my life looks like right now. And then of course you have to have, you know, social network outside of your work life. Sometimes it can blend in, but you know, social network in life to support you as well. So, you know, right now family is really, really important. And basically when I’m not at work, I’m spending time with my toddler and her new sister, who’s gonna be coming soon, and spending time with, you know, family, pets, whatever you can find to find balance and perspective in life. So, anyway, so with that, Happy to discuss any of this, but I mean, things are just crazy and a lot of my work now gets done, in the hours of like 10 p.m. and 1 a.m., but it’s worth it. I think ultimately, there are definite challenges to having or trying to do it all and you can’t do it all well. Just that you have to shift your priorities according to the needs. So with that, I would like to now shift gears and talk about research and this, what we’ve been doing, one of the main projects in my lab, which is to use adaptive deep brain stimulation to treat gait disorders in Parkinson’s disease. And these are my disclosures and these are the relevant grant support for this study. So as many of you know, gait and balance impairments are a major cause of morbidity in people with Parkinson’s disease. And that may manifest as difficulty in initiating walking, maintaining the reciprocal rhythmic locomotion patterns between the two legs, or adjusting body position in response to perturbations in the environment. And unlike some of the appendicular symptoms of Parkinson’s disease, like tremor, rigidity, bradykinesia, advanced gait disorders, such as freezing of gait or loss of postural control actually do not typically respond that well to DBS therapy unless they advance, they stop responding altogether. So one of my main research questions is how can we use neuromodulation to treat these advanced gait disorders in Parkinson’s disease? So one of the challenges to treating gait disorders in PD is the fact that the neuro control of gait in humans it’s actually not very well understood. Human gait, if you think about it, is a very complex motor task, and we’re actually the only species that maintains upright bipedal walking for all of our lives, essentially. It requires a precise coordination of movements across multiple limbs of the body while maintaining balance. And during continuous walking, each leg alternates between the stance phase, when the foot is in contact with the ground, in the swing phase when the foot is in the air. And due to a lot of methodological constraints, actually has been really difficult and challenging to study brain network changes during these dynamic movements and what controls our balance and walking is very, truly not well known until recently. So as Stephanie mentioned, recent technological advances can allow us to study human gait in a more naturalistic way, using completely implantable devices. Yeah, these bidirectional neural interfaces, you know, the summit RC plus S system is also what I use in my lab. And from this, you know, we can record neuroactivity, these local field potentials from, you know, deep within the basal ganglia, as well as from the cortical surface. And these neural signals and LFPs, field potentials can be streamed wirelessly to a tablet. And this can allow us to, you know, synchronize these, bring activity to natural walking in a relatively artifact-free type of way. And using this method and the system, now we can really begin to understand the human neurophysiology of gait and gait control. So using these devices, we have recently shown, this is work led by my very talented postdoc, Kim Lui in the lab, that the septalamic nucleus and the primary motor cortex field potential show dynamic changes that are synchronized with the gait cycle. Again, this is the diagram of the gait cycle, which I’ll be referring to a lot. We’ll just start with heel strike, then the right toe off, and then you go into the right leg swing phase, and then followed by right heel strike, left toe off. And then basically that ends with the heel strike. So this is one full gait cycle. So in three patients implanted, you know, with bilateral ST and DBS, as well as M1 and S1 motor cortical and primary sensory cortical paddles attached to the RC plus device, we characterize their gait, natural gait patterns while they’re on medication. And these are patients who can walk normally. So they don’t actually have any of these advanced gait disorders. And we’re just trying to see, you know, in as close to a normal state as possible, whether there are dynamic changes that occur. And this is just an example of a grand average spectrogram. So aligned to a full gait cycle for all three patients actually. And you can see during the weight acceptance when it’s a lateral leg to the contralateral leg. So this is the left brain, this is the right brain. During this weight transference, you can see strong synchronization across a lot of these low frequency power bands from theta all the way to low gamma, high beta. And this is actually reflected in coherence with the motor cortex as well. So what this, what, you know, Ken found and what this data suggests is that low frequency oscillation changes throughout the gait cycle. And this may play in a really important role in organizing reciprocal muscle activation across the different limbs during walking. So how does this all change in the context of Parkinson’s disease and current DBS therapy? So as many of you know, Parkinson’s disease is now considered a network disorder or a circuitopathy. And some of the symptoms and signs of Parkinson’s disease is caused by excessive synchronized pathological oscillation within structures of the basal ganglia or between structures of the motor network between the basal ganglia and motor cortex, for instance. And one of the key mechanism of DBS is that it can kind of desynchronize or decrease these pathological synchronizations to reduce these motor symptoms and signs of dystonia, tremor, bradykinesia, dyskinesia, just as Stephanie described earlier. And again, this is one of the hypotheses of this to why DBS can work to reduce these motor symptoms. But there are also a lot of challenges with current DBS therapy. And again, you know, traditional therapy is continuous, high frequency stimulation. It doesn’t change based on patients’ movement states. And while this can work pretty well to reduce symptoms of rigidity, bradykinesia, as you’ve heard, just heard, dyskinesia, these symptoms more fluctuate on a larger time scale. So in treating somebody who have these symptoms kind of despite best medical therapy, PBS or continuous stimulation can work really well. But in treating something like walking and gait, it probably doesn’t work quite as well since they can’t flexibly change stimulation settings to match the dynamic neural process that occurs within the brain during walking. And then another limitation of chronic suppression of oscillatory power coherence within the motor network is that it may actually disrupt some normal information flow between areas of the network. So, you know, transient increases in coherence or in oscillatory power is really important as you know what Ken has shown, and probably organizing a complex motor task like walking. So given these challenges with current stimulation protocol using continuous DBS, our goal is to design a smarter adaptive neural stimulation for gait that takes advantage of the real-time sensing capabilities and the built-in onboard control states and algorithms on board the device. So we can alter and adjust DBS simulation parameter to restore natural gait patterns. So in order to do that, first we need to understand the neural control of gait, actually describe it on a larger timescale, how that responds to changes in medication state, changes in response to continuous stimulation. Then we have to identify signals associated with the gait cycle. So swing phase, double support phase, stance phase, and also what’s normal and pathological in the gait patterns. And then we can change stimulation parameters in order to hopefully restore the natural patterns of walking. So, I have my own FDA IDE to use SMRC+S device. And since our approval back in 2021, now we have fully enrolled six patients for our pilot study. And these are six patients with Parkinson’s disease and they qualify for deep brain stimulation due to their motor fluctuations. The other thing for these patients is that, in addition to their motor fluctuation, one of the things is that they have off medication gait and balance disorders, which gave us a really nice way to study what’s more normal or more effective gait in these patients versus if we take them off and that’s what are disordered and pathological gait patterns. And today I’m going to focus on patient two and four, and we actually have successfully tested adaptive DBS for continuous walking in three patients so far. And these are kind of the reconstructions of their locations of their DBS leads, placed in the pallidum, so involving both internal, external pallidum as well, and the cortical contacts are spanning mostly M1 and with the contacts also spanning pre-motor as well, bilaterally. And how we perform this study is that we have patients where kinematic measurements used, we use inertial measurement units, IMUs, to capture their full body kinematics. We also use force-sensitive resistors and gonometers capture heel strike, toe off to really capture different phases of the gait cycle. And we’re able to synchronize the data streams with our bilateral RC+S. And this is just an example of a recording using the RC+S built-in accelerometer. We can synchronize that with any of our external devices during walking. And the first thing we did is actually characterize the neural signature of gait in all the patients in both the low medication state when they have disordered walking problems and on medication state. So this is an example from one of our patients and these are, you know, single trial basically capturing a full gait cycle from left heel strike to heel strike and these are other gait events that occur during the gait cycle. In the low medication state you can see that basically the step cycle length is highly variable and the timing of these gait events is not very synchronized. So this patient, you know, has abnormal walking and when you’re looking at the palatal alpha power and average M1 beta power, just taking two power bands, you can see that when the patient’s actually in the on-medication state, when the walking becomes more symmetric and more more organized and less varied, there’s actually strong synchronization, for instance, in the double support period from the pallet of alpha and during the swing phase of the contralateral leg from M1 beta. And so basically for each patient, we describe, you know, what their brain is doing and what the different oscillatory frequencies are doing during their gait cycle. And using this information, we’re able to, or Ken is able to find biomarkers for the gait events. And here is kind of our scheme for developing adaptive DBS. So first, you know, we record patients walking at home and also in the laboratory. And we identify different biomarkers related to different phases of their gait cycle. And then one thing we want to change is altered stimulation amplitude actually in accordance to the gait cycle. So this is an example of kind of the LDA, the linear discriminant analysis state classifier that we use to alter stimulation based on the swing phase. So right now we’re trying to target contralateral leg swing phase of the gait cycle. So in the upper panel here you can see the actual time points of when the patient is taking left swing or right leg swing. And then using the left brain LDA, we’re using the left GPI in here as an example biomarker, we can track phases of right leg swing based on the amplitude of this biomarker. And basically when this left GPI low frequency power reaches above a threshold, it changes to a state one in which we try to rapidly ramp up stimulation, for instance, or ramping down stimulation in this instance. And so this is kind of the paradigm we’re trying out in our patient. So here’s an example of a real-time implementation of adaptive DBS during walking. So here is kind of the X-SENS inertial skeleton of the patient. So you can see him walk, taking steps, and he’s basically just walking back and forth along a straight line, unobstructed, at his own natural pace. So the upper strip again is the actual gait phase, left versus right leg swing. And here for simplicity’s sake, we’re just showing the right GPI spectrogram. Again from 5 to 35 hertz, the dotted lines show the frequency range in which Ken has identified to track well with this patient’s right leg swing. So basically, this biomarker, which is in the low beta frequency range, seems to increase during or just right prior to right leg swing. And then they decrease during left leg swing. And the bottom is the classifier, this LDA state detector. So basically, we set state 0 to be when the right GPI beta power is high. So that should be the right swing phase. And then one to be when the beta power is low. So that should be the contralateral, so left leg swing phase. So basically the end result is that we’re rapidly ramping up and down stimulation between, you know, 2.5 milliamps and four milliamps, which is kind of the patient’s clinical continuous stimulation amplitude. So we’re trying to increase stimulation during left leg swing, so contralateral leg swing. All right, so here’s an example for patient walking at slow down. And you can see again, when there’s a rise in this beta power, which seemed to correlate pretty well with the right leg swing phase, and we decrease the power. But then during the left leg swing phase, it seems to track fairly well. And what our team would do is, you know, do this program this for the other side of the brain as well. And here’s a plot of, you know, how accurate actually implementing this, you know, rapid on and off stimulation based on just neural biomarker is. So in this one, we have tested this adaptive, different adaptive DBS setting in this patient three times now. And we used, you know, both the Paladin GPI as well as the premotor M1 biomarker for detection for each hemisphere. So the upper panel kind of shows left hemisphere accuracy and then the bottom shows right hemisphere accuracy. And you can see the solid color shows overall accuracy. So overall accuracy is a comparison of the state at each time point during the whole recording session and what state it should be in, whether it is in the left versus right swing phase, for instance. And then the phase accuracy looks at weather stimulation parameters. it’s actually changed during the specific gait phase that we’re trying to target. So out of, for instance, all the left contralateral leg swing cycles, how many of those were we able to actually change the stimulation parameter? So ideally, you want high numbers for both, but you kind of had to look at both together to see the overall accuracy. And in the latest testing session, So Ken’s and our team tested five different adaptive stimulation settings, and we’re targeting different phases of the gait cycle. So the green here, settings one, two, and four, we’re targeting contralateral leg swing, and then setting one’s based on motor cortex and premotor cortex biomarkers. And then the two and four are two different GPI biomarker settings. And then settings three and six are identical. trying to target the double support phase based on premotor cortex biomarker. And from this, you can see that it’s not ideal yet, but at least we’re able to rapidly alter stimulation, which is not an easy feat in this system. And just give you an idea of what this looks like in real life. Again, the top is actually a patient’s continuous clinical DBS setting.

and we can see that his left leg it’s a little bit slow in taking a step and a lot of times you know step length is not quite as good as his right leg but down here and again these videos are shown at the right speed for the same amount of time he’s actually walking a lot faster in the ADBS setting. We’re using here the GPI as a biomarker for contralateral leg swing. And when we characterize the actual gait parameters, this is looking at step length, which is the distance between the point of initial contact of one foot and the point of initial contacts so heel strike of the opposite foot. Again, setting five is this clinical DBS setting and the patient and most of the examiners except for Ken is actually blinded during these testing sessions. You can see again his left step length is a little you know shorter than the right and using the adaptive DBS setting and we have actually a physical therapist Jessica Bath who’s a graduate student in my lab. So she’s also blinded, you know, she observes the patient walking during these test sessions. And she thought, you know, maybe setting six was the clinician preferred setting for the patient. You can see that the step links are definitely a lot more symmetrical. And this one is the one which patient felt the best. So he on prompted, told us, you know, the, his walking felt faster and more fluid. So interestingly, you know, both clinician and patient prefer ADBS setting compared to continuous traditional clinically optimized DBS. And then also certain ADBS settings can increase gait speed. So this is stride time, which is basically the duration of a full gait cycle. And this is for the left side, right side. And again, this was a clinician preferred one. So she definitely noticed an improvement in gait speed using this double support paradigm. And then patient himself also picked the second best in terms of gait speed. So, you know, this is all really promising. The question is whether we can recapitulate this in other patients. So, you know, we also did, I think, two adaptive DBS testing sessions with our patient number four. And this is just showing the first one. I think the hot off the press testing session that my group did two days ago showed even more promising results. So here, you know, again, we just targeted contralateral leg swing based on a variety of premotor, motor cortex, as well as palatal biomarkers. And for him, you know the accuracy is mediocre it’s not really significant or that much better. We’re capturing maybe about half the gait cycles in terms of changing stimulation parameter during the contralateral leg swing and this is you know his gait parameters. So he also has you know shorter left leg swing or left leg steps compared to right leg steps using his chronic or continuous DBS setting. And you know in some of the settings we tested, again this is the clinician preferred, this is patient preferred, we make you know the step length a little bit more symmetrical. But you know the right leg step length actually shortened a little bit. But you know certain, again, certain ADBS settings may be able to improve symmetry and step length in this patient. And in terms of gait speed, you know, he was walking pretty well before. We can alter, you know, his symmetry and step length without compromising gait speed. So, you know, again, patient tolerated this extremely well. He didn’t, you know, have many major complaints. And it seems that, you know, at least we got it to work. And what’s interesting in this patient is that, you know, besides tracking kind of gait phase, which we’re so optimizing on, we actually kind of inadvertently maybe found a signal or biomarker for when he’s taking a turn. So again, these are, you know, left brain, right brain spectrogram, this shows kind of the, the band that we’re tracking. And I want to lead you to this kind of high beta, like 23.5 to 27 narrowband frequency, which seems to increase in power, you know, really right before turns in this patient. And so in a lot of patients, you know, turns requires a lot of these anticipatory posture adjustments, and that’s when people tend to freeze or have trouble. So if we can, you know, really detect a good biomarker for turn, and that’s when we turn on these ADBS settings and change the stimulation parameters, that may be actually really beneficial for some certain patients. So, you know, this is all really nice and all, you know, in the clinic, our ultimate goal is to embed all these algorithms at home. So people can go home. So when they’re walking, they can switch to kind of this adaptive DBS that matches their gait patterns. But when they’re sitting, not doing anything else, it’s probably better that they switch to a different setting, maybe a clinically optimized setting to treat their tremor or rigidity for their arm symptoms. So in this project, we also track patients walking at home using this external wearable device. It’s called the Rover Health. Basically, it’s our ankle bracelet accelerometers that can, with a built-in algorithm that can detect periods of walking, showing green here, versus when the device, the onboard algorithm thinks the patient’s not walking. And this work is done by Rithvik Ramesh, who is a medical student who worked with us for the summer, and my postdoc Hamid Asghomi. So in one particular patient, we looked at over 10 hours of home recording and we saw over 1000 gait events and over 2000 non-gait events that’s classified. And we just take different average spectral power during about 10 seconds of power. So we again, stream their neural RCS data. while they’re either sitting or walking. And as you can see, based on the rover classification of walking, there’s very different spectral profile in that Nangate event has overall increased alpha beta power compared with walking events. And then based on this, just using their brain signals, Rithvik developed a kind of a, again, a simple classifier, LDA just to classify walking versus non-walking periods using the brain signatures. So this is an accelerometer from the RC+S device. And again, these kind of these periodic episodes indicates walking. And basically he’s just using the average power bands to classify walking versus non-walking events. And it’s the very first pass, but it seems to work reasonably well in terms of classifying periods of walking and non-walking. And again, we’re using 10 seconds of average data here. And we have patients do certain protocols at home. So these kind of continuous episodes is probably a patient walking back and forth. So the periods of low acceleration, he’s probably still walking, but just turning. And again, using a different combination of regions. So key one and two are the basal ganglia electrodes. Key, actually key zero and one are basal ganglia, and then two and three are cortical contacts. He’s able to get pretty good AUC, so about 70%. And if he’s using multiple sources, he can get up to 80% accuracy into detecting and classifying walking versus not walking periods at home. So this will kind of give us additional specificity in employing our ADBS FAST algorithm at home. at home. So in conclusion, what we’ve learned so far is that both the motor cortical areas and basal ganglia demonstrate oscillatory changes that dynamically change during specific phases of the gait cycle. And we can use these biomarkers of gait phases to control timing of stimulation. So you know, one thing that we’re trying to do is that normally stimulation is good, but too much of it may be a bad thing. So timing of stimulation is really, really important in helping with these kind of dynamic movements in Parkinson’s disease. So by altering and controlling timing of stimulation synchronized to the gait cycle or intent to move and intent to walk, this may help us to restore kind of the natural neural changes that occur during the gait cycle to improve gait functions in patients with Parkinson’s disease. So next we plan to test ADBS algorithms in the clinic with all six of our patients and then we plan to employ a blinded trial to test conventional continuous DBS versus ADBS at home and then track their gait measurements using the home wearable devices. So with that, I would like to thank you know members of my very talented and hard-working lab without whom this none of this work would have been possible. And the people the group that’s really involved with the GATE work include Jenny who’s our amazing clinical research coordinator, Jessica a graduate student I mentioned also our physical therapist, Hamid, Ken, Rithvik, and who all contributed greatly to this work. From Medtronic for providing our patients and access to these devices, our patients, our collaborators and funding sources. All right, and with that, I’ll take any questions. Amazing, thank you so much. Thank you also so much for your openness and sharing not only the beauty, but also the challenges of juggling these different hats that you’re wearing in your daily life between being a neurosurgeon, scientist, and also a mother. We now have time for a shared discussion. And I already saw a couple of questions incoming in the Q&A chat. But if there are any live questions, please just raise your hands and we will then unmute you and you can ask your question live within the session. Maybe I can steal the privilege of asking the first question to Stephanie because we’ve heard a lot about the passion that Doris has for her work. And as an icebreaker question, I thought maybe to ask you what most fascinated you about this trial that you presented and if there was any surprising moment or eureka moment of sorts that you experienced.

Yeah, that’s a great question. I’ve been interested in adaptive or closed-loop stimulation since my PhD. I had the opportunity to work with Bishop Ogundas at the University of Florida, so I’ve always been interested in it and kind of implementing it at home because my PhD work was in the clinic. So it’s been amazing to actually implement adaptive at home and see that it works. I guess the eureka moment is I don’t think I ever expected finely tuned gamma to come out as the most distinguishable control signal. And again, it’s only three patients, but that’s exciting. And I think it kind of opens the field of adaptive stimulation to kind of look beyond beta since that’s been kind of the key control signal. >> Really cool. And what did you expect instead? I honestly thought it would have been beta and it was exciting that it wasn’t. Yeah, I thought it was just going to be beta as everyone has been using, but I think bringing in a new control signal kind of opens everyone’s minds to that it can be across the frequency abandoned to not solely focus on one area. Cool, thank you. We don’t have any raised hands yet, but we do have a lot of Q&A questions, a lot of interest in your talks. Garance, do you already want to start reading out some of them? Yeah, so we can start with a question from Jésus Sain, which was, “So I’m a PhD student and I have a question for Stephanie. Could you expand a bit more about all that can be common biomarkers when the most bothersome symptom for each patient may be different? Thank you for the great talk. Could you hear it because there was some noise? Okay, so the question is, could you expand a bit more about how there can be common biomarkers when the most bothersome symptom for each patient can be different? Yeah, that’s a great question. I think in this case we were really tracking medication state because their most bothersome symptom was really an off medication symptom. So once we had a biomarker, I think myelin 2 and gamma was really representative of the on medication state and that’s kind of how they all aligned across all of these different symptoms. Okay, so if I understand correctly, you think it’s more belts. So it’s a common marker because it’s more indicative of being off medication. And if, for example, the most bothersome symptom would be more dopaminergic resistant, like maybe tremor, it wouldn’t be well reflected in this marker? >> Yeah, I think finally tuned gamma is really indicative of the on medication state, so representing their hyperkinetic states. So that’s why we didn’t really see it.

I don’t know if you remember the spectrogram I showed, but when the patient marked their most bothersome symptom, there was an absence of finely tuned gamma. So that’s when we really increased stimulation amplitude. But if someone had tremor, exactly, I’m not sure how this would work, or if finely tuned gamma would pop up. We are doing another in-clinic study with these patients to kind of probe finely tuned gamma a little bit more.

and we’re switching stimulation frequencies, we’re tracking it as it goes to half stimulation frequency and how that’s related to symptoms. But yeah, I’m not sure since no one in this study really had tremor in the off medication state what the biomarker would be, but we would be open to them all. >> Thank you. So we have several questions from AIDS Goon News. Do you want to, I saw that you had your hand raised, that sometimes you want to come live if you’re still here. Yeah, you’re allowed to talk. I think. All right, sorry for hugging the questions, but again, so the fine gamma appears at half the stimulation frequency, which is a therapeutic stimulation. And I think April had asked, what are the placebo stimulation tests or even the control stimulation? So do you think if you were giving a non-therapeutic stimulation frequency, and let’s say it’s really low, like 60 Hertz, and obviously if it’s half the frequency, first of all, do you think we would see a peak there? And then, obviously, being 30 minutes, it might also interrupt the beta-based classification. So that’s a really great question. Myself and Maria Shcherbakova– I showed her on the first slide really quickly. She’s our clinical research coordinator/software engineer. And we’ve been looking into what happens when we change to other stimulation frequencies. So sneak peek, we do see beta in trainment and we’re working on relating that to, is that related to an increase in bradykinesia? ‘Cause now we’re in training beta bands. So it’s all upcoming. But yeah, we see it switching to half stimulation frequencies at both therapeutic and non-therapeutic. All right. Then a loaded question, ’cause you still owe me a paper. How would the effects, or how would we assess side effect mitigation with close-up stimulation in Parkinson’s disease? So in this case, we just asked them, so for side effects in our three patients, we asked them about the time spent with these symptoms at home. Do you mean objective assessments?

Yeah, it could be objective, but especially the side effects. Obviously, dyskinesia is now a side effect, but we also say, you know, continuous DBS has general side effects. I’m not as familiar with all the side effects in PD, so in that case, if you could say one side effect, maybe Doris can Also chime in. And so, yes. In our case for dysarthria, since one of our patients did have that as a stim-induced side effect, the next phase when we’re doing the two-month blinded trial, we are doing objective assessments of speech in the off and on medication states, both on continuous and adaptive stimulation, and look at different things of speech, especially intelligibility, to see if adaptive helps or not. All right, finally, Doris, and this was when you showed the first figure, so you know, you did talk a lot about both the STN and the cortex, but for the gait cycle, there were some, you know, opposite trends. So how does that affect the coherence? Right, so we saw coherence. It’s more variable. So STN showed a more clear pattern, you know, during like that double support period of weight acceptance of the ipsilateral leg, we see this clear increased SDN power. And during the coherence between STN and motor cortex in patients, it’s more smeared. So it kind of bleeds from that, you know, weight acceptance, double support period, into the beginning of the swing phase, contralateral swing phase. And I suspect because our motor cortical electrodes are placed kind of more over the upper extremity area, so that might be more associated with arm swing. So we’re probably picking up, you know, parts of arm swing with our cortical contact as well. So in those three SDM patients, it was somewhat variable. In GPI patients, it was a lot more variable, the coherence, just because these patients do have very, as you saw in at least, you know, two of the patients I demonstrated today, asymmetric, you know, step length and arm swing.

All right, perfect. With that I will shut up. But thank you both. So I think I think we had one live question, but the hand just disappeared again. Melissa, do you want to still answer to ask you a question? Or was it maybe already answered? I guess probably was already answered by what you just said.

Maybe I can go with another question. Maybe both of you can answer it. Yeah, maybe most suitable for Stephanie. So the algorithms you showed are quite powerful, but they probably require a lot of expertise as well. So I was wondering how feasible you would think it is to actually streamline them into clinical settings, how scalable they would be?

That’s a great question. Actually, there’s graduate students in our lab now working on making this pipeline completely streamlined. And then essentially, how do we adapt adaptive, because these settings will probably change over time, patients will progress. So we’re currently working on that in the lab as well. We’re getting there. But I think it’s definitely feasible. Very cool. Thank you. You have any thoughts on that, Doris, as well? Yeah, I mean, right now, you know, what you saw is hundreds and thousands of hours poured in from people dedicated to analyzing these patients’ data. But, you know, what’s kind of, I guess, pretty positive, and I think from a patient’s perspective, is that in both the patients I I showed today, it’s actually their palatal biomarker. So we don’t need to, that, you know, actually predicts the best gait phase. So from the patient’s perspective, it’s nice that we don’t have to put in extra contacts and electrodes in their brain. So it might be feasible to do all this based on just their DBS lead. So that will cut down, you know, the amount of data that someone has to analyze by quite a bit. And once we streamline this, I think now with AI and other advances, we might be able to teach a computer to kind of just feed in all the data that we have and then for it to analyze and find these biomarkers in an efficient fashion to put this out in the real world in a greater scale. Thanks. Yeah, as I understood, there’s a lot of patient interaction required at the moment, right? Also, how did your patients perceive going through this procedure of fine tuning or finding the right biomarker? Yeah, you know, they’re the true heroes, I think, for both of our studies by being tolerant with us, by being super, super patient with our research team and all the different protocols. And as you can see, it’s very onerous on the patients to do all these home recordings as well. And, but I think, you know, we select for the right patients. They understand that this is, you know, contributing to science. They themselves may get some benefit from it. And, you know, based on Stephanie’s work and our initial results, it’s very possible that this is better than their traditional DBS. So that’s, you know, some reward that we can give them at the end of it all. But yes, they, having them and supportive partners that they have to help us, helps us, you know, to find what we can find. And they’re really just as much, you know, an important part of the study. I have a related question. So we were discussing how adaptive stimulation can help. So, so far you’re, we’re adapting the amplitude of the stimulation depending on different markers, so on and off state or gate phase. And can you comment on maybe, so it’s always using the same contact, right? What do you think about switching between different contacts and programs? Do you think it could add a lot on top of that or do you think it’s really more about stimulating at the right moment? Yeah, right. It’s because there’s so many different parameter spaces you can explore. And the goal is that right now, they are optimized on their programming based on like a lot of other PD symptoms that they have. I think three or four patients of mine have pretty bad tremor, rigidity. So we’re not trying to reinvent the wheel right now, at least like exploring too much of different contacts and where to stimulate. And also setting it up, they may notice more noticeable change, right? Instead of just decreasing stimulation. So right now we’re just focused on timing of stimulation. The other aspect of the work I haven’t mentioned is actually frequency of stimulation. So there’s actually a lot of work being done by Hamid on stimulation parameters like pulse width and frequency of stimulation. So we’re trying like different high versus low frequency to see if that can help with patient’s gait. So in those patients, again, if this kind of rapid ramping on and off is not as accurate or if they have side effects, what we can do is switch them to a gait optimized stimulation parameter. And he’s using machine learning algorithms and objective measures to see what the effect of different frequency of stimulation has on the brain neurophysiology and gait outcome. And so that’s another way of employing ADBS for gait. Thank you. Amazing, we have a question by Melissa and you should be allowed to speak up now. Okay, hi, I’m Melissa. I’m a PGY5 neurosurgery resident here at the Brigham. I just really wanted to thank Dr. Wang for sharing her journey. Someone who’s interested in functional neurosurgery And I actually was curious, given that you’ve seen this biomarker in GPI, do you think this could potentially kind of shift the paradigm for where to place the DBS if a patient pre-ops says freezing of gait is really my big problem, even though you’ve recommended ST and DBS, or maybe in the future we could even place four leads to an ST and two in GPI? Yeah, that’s a great question, Alyssa. And we kind of have to pick a target given it’s a small clinical study. And for what it’s worth in the patient’s population, at least we see here, usually those with more advanced gait problems like off-medication freezing, posture instability, also have some mild cognitive impairments. So they tend to be patients selected for GPI, DBS anyways, at least at UCSF. So, you know, that’s kind of where we started. But I think there are a lot of other groups in Europe and also at Stanford, I think Helen Bronte Stortz’s group is exploring, you know, STN-DBS, adaptive STN-DBS, the freezing of gait. So, and there could be other targets like the PPN or some other, you know, basal ganglia target we haven’t even explored. So yeah, you know, I think if this is positive, at least, you know, GPI stimulation doesn’t cause significant issues. For people who skate is a primary concern, this may shift, you know, our paradigm. Of course, we had to validate this in the larger clinical study, that they may be better candidates to get palatal stem and implantation rather than STN. Thank you so much. Pretty cool. Maybe we can take one or ask one last question very close because we want to be mindful of your time, of course. So I was wondering, do you see any applicability in the long term of adaptive deep brain simulation also to neuropsychiatric applications and what will be challenges or factors to consider in these cases? Do you have any guesses? Probably not so much your focus of research, but interesting to hear what you think about that. >> Sure. I think it’s definitely going to be tough because in, I guess in our case, we can quickly see changes in these symptoms. For a neuropsychiatric, they tend to happen over a long course of time. So that’ll definitely be challenging. And then it’s also, would respond better to adaptive stim or would something like, I can’t think of the word right now, but essentially turning stim on for a long amount of time and then turning it off for a long amount of time be better for them versus adaptive? I think it’s doable. I just feel like there still needs to be a lot more work done towards that and it will be difficult to kind of capture the changes due to adaptive stimulation. But I’m excited to see where the field goes. Yeah, I think like Stephanie said, there needs to be greater understanding of actually the network physiology changes that occur in neuropsychiatric disorder. And even in PD, right, you think it’s a well-defined disorder. There are so many different subtypes, like the patients enrolled in, you know, Phil studies and Stephanie studies, they have different set of symptoms than the ones in my study. And those are with objective, you know, measurable outcomes. So in psychiatric disease, like depression, OCD, for instance, there’s, it just gets a lot more complex. But I think there are a lot of groups who have made significant strides using these devices for long term study of network dynamics that contribute to these symptoms, either positive or negative symptoms of neuropsychiatric diseases. So at UCSF, we have the Presidio trial where they implant multiple electrodes in orbital frontal, a lot of patients in different locations, hundreds of contacts, and they undergo stimulation in the inpatient setting before they decide which contacts and electrodes to use for the outpatient setting. And you’ve seen probably some of these reports of positive results and then Sameer shot that email. Baylor has been looking at OCD for instance. So we’re like delving into it and I think it’s going to be, there’s definitely a role I think for ADBS and at least the sensing part to understand the disease better and I do think ADBS and neuromodulation in general would be a treatment that probably will become, you know, FDA approved and kind of one of the major goal standard of treatment for medically refractory neuropsychiatric disease in the future. >> Really cool, exciting outlook. Thank you so much. I’ll hand over to you to do the honors of closing. >> Yeah. Yeah. If we don’t have more questions, I think we can close the session. So thank you very much for participating. It was an amazing session. Amazing to see different aspects of adaptive DDS and how it can be used for different symptoms and in different ways. Our next session will be on September 28th, sorry, and it will be about, so we will have something about tremor and about quality of life after the brain stimulation surgery with Heda of DAF3 and Gunther Doshu. So we hope to see you there. Thank you to the attendees. Thank you so much. You have a great day. Thank you. Bye.