Jin Hyung Lee: How can we systematically cure brain diseases?

Jin Hyung Lee: How can we systematically cure brain diseases?

Show Video

today on the future of everything the future of brain engineering so brain diseases are among the least understood diseases in medicine and are often difficult to diagnose and they're almost always very difficult to treat and in fact for many brain diseases we do not have effective treatments part of this is because it is clearly the most complex organ in our bodies and it has been very difficult to study and to learn the principles of how things go wrong and what we can do to correct them now there have been remarkable advances in understanding the brain the us national institutes of health has a brain project which is a 10 year project to really try to advance science and it's devoted to increasing both our understanding of how the brain works and the creation of tools to measure brain function this will hopefully lead of course to new treatments now we are able to do some things really well many of us know about ct and mri where we can take pictures of the inside of the brain we can see the structure and the tissues these pictures are becoming increasingly detailed and cannot even show the details of which regions connect to which other regions it's kind of giving us a basic wiring diagram which is not complete but we're getting there we can also measure electrical activity many people have heard of eegs where we measure the electronic waves in the brain to understand the complex patterns associated with both health and disease and then we have increasingly these functional measurements they're called where we look at where the brain is working where is it using glucose or oxygen and that's a clue to when we do certain activities which parts of the brain are doing all the work so things are promising but we're not there yet we have many challenges we have alzheimer's disease we have strokes we have seizure disorders and of course many inflammatory disorders of the brain well dr jin lee is a professor of neurology neuroradiology bioengineering and electrical engineering at stanford university starting as an electrical engineer she has dove into neuroscience and is using technologies and approaches of engineering to study brain function and disease gin you have said that there has not been enough of an engineering approach to the brain and brain disorders what has the previous approach been and how are you trying to change this thank you russ uh it's a great pleasure to talk about the future of brain disorder treatment from an electrical engineer's perspective and i think one of the biggest challenges in brain disorder treatment is that we haven't really had a systematic approach to do things i mean we have worked through itunes space we all of us like no matter what age we are we've seen things change over the past few decades where it has gone and taken over our life i.t as in information technology yes revolutionized everything that we see however when you just take a little turn and look at the space of biology things change dramatically where you are looking at things moving at a really sluggish pace and once you have things like brain disorder if your loved one has a brain disorder you come and look at it what are your options basically nothing and so why is that one of the key reasons why we could revolutionize information technology is because we had a platform to set goals design the desired outcomes and execute but in biology medicine for the most part it has been done through hypothesis-based research yeah so this is kind of the way science happens you have an idea and you see if it's true and i know that you've said that that's not ideal so explain why that isn't a good way to go anymore or at least it's not enough it's a very important start to try to understand something but that doesn't really cut it in a sense that you know you have descriptions of something it's sleeping is better for your brain how does that help you figure out how to fix something like you have alzheimer's disease and where do you start there uh okay i hear that there is more inflammation in the brain when you have alzheimer's reducing inflammation okay maybe that helps maybe that doesn't that doesn't necessarily help you figure out set the goals of your problems which makes it very difficult to solve ideally what you would need to solve problems like that is to say okay this is what the normal brain function algorithmic execution looks like this is what a brain that has suffering from alzheimer's looks like then you have a very clear set goal is whether it is through having inflammation reduction whether it is through having uh plaque reduction those are all things related to variables you want to control in an engineering problem you have to be able to set your goals very clearly yes and that is not possible to date because of various reasons but also there is very little effort to set the goal um many are just approaching it from more of a biological hypothesis driven so is it the idea that that gives us only a patchy understanding of like we have a little knowledge about this and a little it's like almost like swiss cheese but then when you're trying to do like diagnosis of diseases you need a much kind of broader look at the at the disease and what and how to intervene right so if you use an analogy with like uh i a phone that we have we all carry one imagine you took the phone to the iphone to an apple store and the genius tells you did you drop the phone did you like um put it in a hot place or did it go into water those are all important clues about what might have gone wrong with it but if that's the only thing they knew about how to fix it you're not facing a very positive okay so that so that apple brain person is like today's doctors asking you about your symptoms and saying well based on all this stuff i think you have alzheimer's you would like to have a little bit more of an objective it sounds like way to assess the brain you need to be able to pull out the tools diagnostic setting run you see oh this is the part where your software is not running properly you have to be able to look at oh this component is broken and hands reducing your function here and there one thing i want to say very clearly is that the brain treatment what is the goal of a brain disease treatment what do you think well okay you put me on the spot but i would say that we would like to return to uh previous or normal function where whatever normal means to that patient exactly so what you need is to be able to define the normal function let's say you have a parkinson's and you're shaking you need to have a normal range of brain function that controls motor system and then you need to know what happens when you're shaking and define the circuitry that is malfunctioning as a result of that and once you have that goal set of restoring normal brain function you need to be able to do all kinds of variable testing you have this drug that lowers your plaque accumulation you have this drug that reduces your immune system response everything you have to attack this issue of it's like drying the phone or you know replacing some components all of that main goal is to restore the function of this phone that you has to do various functions and for the brain it's the same thing you need to be able to define the function to an extent that you can reproduce the behavior that you're interested in if you can tell how the motor control works to have normal function executed then if you can distinguish that in a disease state your goal becomes to simply minimize function normal function disease minus minimize that function and you have various variables to tweak in order to make it happen this is the future of everything i'm russ altman i'm speaking with professor jin lee about engineering brain disease so um i think you you've been pretty very clear that we need to be able to make it sounds like quantitative measurements and then we need a very large tool kit of interventions so that once we know where things have gone wrong in the disease we can turn some knobs with these interventions to try to get things back so i in my introduction i talked about some ability to measure that we can take pictures with mri we can measure electricity is that enough to start doing the kinds of um i would say this engineering approach to fixing things or do we just simply not have enough ways to measure so that we know how to intervene so i guess my question is i'm sorry i'm talking so much welcome to the show i guess my question is do we need to put more effort in making the measurements or developing the interventions based on the measurements or both i would say both but one thing that is missing right now is one what we've done so far is fantastic we've made all these tools to measure things and measurement is the beginning of everything where you know imagine trying to lose weight while you can't measure your weight so that's my favorite saying from engineering is if you can't measure you can't intervene exactly uh the thing with the brain is we have a lot of different tools to measure things but they're not quite enough what you need to measure is let's say you have alzheimer's disease one of the things um that i i often talk about is if you have if you suspect you have alzheimer's disease you go to the clinic are they able to tell you you have alzheimer's disease it's very difficult unfortunately i have loved ones and basically they do a zillion tests and if none of them come out positive that is another explanation for the dementia then they're left saying well it must be alzheimer's exactly so most of the tests that are used for these neurological disease testing is diagnosis of exclusion figure out oh you have you took an mri there's a brain tumor oh that's that's why you're having these memory problems or things like that they're all diagnosis of exclusion because of the fact that there is nothing that can tell you the algorithmic dysfunction underlying the brain okay there are many measurement tools out there which could start to make those information be accessible but we can't quite as of yet we can't quite give you the information on how to take those measurements and then tell you that oh your memory function is a little bit off your memory function should be working like this but it's working a little bit like that so we have to tweak that back to normal that's what we need but none of the tools that we have in the clinic right now is able to do that now i know that you are working on such tools um so tell us about that what what how how is the lab doing this so the biggest difficulty in trying to figure this out is the brain is a very complex system and i can't open your brain to figure out what's in there and also once i open it it doesn't really do what it's supposed to so it's difficult to figure that out but there are many tools that are now available in the labs and there's a little bit more things that you can do with animal experimentation and so we are now taking these experimental tools and we have looked at how we can reconstruct the algorithms underlying specific behaviors if you have parkinson's related repetitive motion when you have that motion what system is executing what function to make that happen when you have epilepsy where uh certain brain signals are malfunctioning and creating these synchronous states that it shouldn't have what exactly is generating that and maintaining that system in this context of a brain circuit so we now have these measurements that allow us to reconstruct these brain functions underlying the behavior of interest and once we have that we can now take this and look at clinical data to reconstruct what it means basically i can't necessarily take detailed enough pictures of your brain so are these mostly mri type measurements it is a combination of using genetic tools and mri imaging and also electrical recordings where we actually take a multi-resolution top-down approach uh we go from uh the large-scale network we look at how specific cell types inter-functionally relate to produce a specific behavior and so we get a large scale map it's like a highway interaction map and then once we have that we also go in and look at a little more detail and see how specific neurons one by one talk to each other to generate that map where we can now construct models of how algorithmic execution led to specific behavior so when you say algorithmic that's interesting because that of course that's kind of sounds like a computer like there's a series of steps so it sounds like your your model is that these diseases have the same kind of underlying logic uh that needs to be fixed like basically it's an algorithm that's gone wrong it used to say go left now it's going right or you know or whatever so is that a good model for these neurological and diseases that's a great question um oftentimes people ask are all our brains the same like because you know clearly your brain and my brain are different in several ways but actually it is the same in the way we control motor function for example your brain and my brain does the same execution code to lift the arm uh it may be we may be thinking about different things but that's another set of things there are common architecture that describes the basic function that we all share as humans and so that part we there is a range of normal so to speak in how things are executed and we have to define that range and then we have to also be able to tell what's different there and when you when we say there's a disease we there is clear differences and the key is to figure out the algorithmic precision level that can distinguish these things and then once we have that our goal is to merge this gap and so we can now we are now actually just starting to be able to construct these algorithmic models of how these behaviors are executed and so with that now that we have these clues from experiments on how these algorithms are executed we then go even though i couldn't measure everything from ross's brain perfectly i have all these different measurements of mri and eg and ct and whatnot and we can then take that and then from that clue we can construct these uh underlying models of what your motor function execution look like for example now when you when you say underlying model um i i i did do my homework i looked at some papers and your group is um increasingly publishing papers where you're using these ai type methods like deep learning and other is that what the model is is it's like is it a computational model of what's going on in the brain or are those ai algorithms more for like unraveling the signal so that you can see what's going on the short answer is the latter the brain is a complex enough system where the ais cannot really figure that out we are really basing on data that's directly measured while certain brain function is executed and then that data is used to construct the model ai aspect is mainly to just clean up data and then uh get good signal out of what we've measured uh it's strictly based on experiments that can produce the behavior direct data translation into models and ai is just used as aid to uh be able to make the data more yeah so that's that actually makes perfect sense so all this mri and ct and eg they all have noise in them and if you can clean up that signal my guess is it's like cleaning up a photograph where instead of a blurry image now you might see the kind of detail you were just speaking about a few minutes ago that give you much more precision and understanding normal versus disease in one particular area of the brain well uh this is the future of everything i'm russ altman i'm speaking with jin lee and we'll be back with more on brain measurements and brain interventions next on siriusxm welcome back to the future of everything i'm russ altman i'm speaking with professor jin lee we were talking about measuring the brain and now i want to kind of move over to intervening you said in your vision for the future of brain engineering you said that once we make these measurements and know what normal light looks like and know what disease looks like then we can start making interventions to try to move the patient from the um towards the normal state so can you give me a sense of what are those interventions going to be is it going to be a shot in my arm is it going to be some funny little hat that i wear that's doing electrical input into my brain paint a picture of the future fantastic so what i what i envision as a future once we have these technologies deployed into the clinics would be where you know you go to the clinic instead of talking to the doctor discussing what you might have instead you will get brain measurements which will look at your brain's status of how everything is functioning if you're suspecting alzheimer's you will measure things relevant to alzheimer's and then look at that brain status and you'll get a score okay you have this particular function a little bit off so you are at the verge of going to the alzheimer's stage and then now you know the problem once you know the problem where you can deploy all the different techniques that we have currently in our uh um toolkit and those toolkits now it's difficult to use those toolkits partly because i don't know what's wrong with you right and if i use toolkit a does it work or not i have to see whether you're you know did it respond yes no come back in three weeks let's see if it worked right but instead what we can have is okay this is exactly what's wrong with you and based on what i know about these different toolkits function i think toolkit c is what would work for you but let me let me press you tell me what those tools might be i know that you don't have a crystal ball but you have the closest crystal ball of anyone i know since you work in this area what might they be that we tool kits are actually visible for a lot of us right now it's just that it's difficult to deploy them properly tool kits would be various drugs that change status right it you know if blood pressure is a problem blood pressure lowering drug even can be something of a toolkit for potential to remove you from going into a disease state inflammation reducing drugs plaque removing drugs or even brain stimulations that are targeted to specific brain regions like electrical yes electrical brain stimulation there are also many genetic toolkits coming where you can edit some genes in order to now you're talking this is what i wanted to hear about so we might be injecting new genes or editing mechanisms many people have heard about the crispr tools for like changing the dna we might say because of your problem we're going to go in and change this little piece of dna to kind of move you towards normal exactly so we have many tool kits now in play the problem is with the brain it's unclear how to deploy those set of tool kits but the kind of research that we're now doing to be able to give you that measure what it allows you to do is even if let's say you do some gene editing you might overdo it and you might push you into the wrong direction right everything about brain disease treatment is to put that brain function in a narrow range of normal right if you cannot measure it you will go up and down even if you find a knob that can really change your brain status it can go into multiple directions also this set of multiple knobs that we have one knob may not be enough you may need to turn three or four knobs to the right range right now that makes really good sense based on what you've been saying about the complexity go ahead i'm sorry but when you don't have a measurement to be able to put it in the range how are you going to tune three or four knobs and so what we are going to be able to see in the future by having this measurement you will individually be able to diagnose the person's problem to give a discrete score of what's going on and based on that score we can deploy all of these fascinating you know technologies we hear about all these advances in biology that allows us to even edit genes right but yet we don't have a cure for any brain disorder which largely relates to the fact that we haven't set a proper engineering goal and we need a tool to do that the other thing that strikes me and it's kind of implied by your comments is you make that initial visit you make that diagnosis and then you give some medications or some editing or you're allowed to make more measurements to say did it work or do i have to dial it back you know in some of the things that i do where we're adjusting medication doses you always have to remember you make your first prescription but then that's why the patient comes back and you tweak things and so i now imagine that you're going to be making your measurements several times over the course of treatment so what about stem cells people there's been a lot about stem cells in neuro in neuroscience is that going to be one of the tools as well is that looking good or is that just more hype and not really a useful set of tools the way i see it is the more knobs you have the better and stem cells is definitely one of the knobs if you have destruction in the brain that's difficult to restore with anything else stem cells is kind of a way to have new component plugged in right so to speak and so for stem cell to be successful i think having a very tight control over the stem cell generation and you so does you need to have a way of making the component very precisely for the purpose that we need and once you have that combined with the ability to assess the status we can say like oh this chip in the brain needs to be replaced and that would be the role of the stem cell the way i envision it so this is the future of everything i'm russ alban i'm speaking with professor jin lee about brain engineering and brain treatments and and you know i want to make this real in the last few minutes we have three and a half minutes i i just want to talk about i know you've published in many areas such as alzheimer's stroke uh seizure disorder epilepsy i just want to get your sense of what's the status of those fields because many of the people listening now either they themselves or loved ones or friends are affected by these diseases so let's start with alzheimer's because it's come up a couple of times before where are we now and what are the big questions or the big engineering um obstacles to having good treatments for alzheimer's so just take a take a step back and tell you about some of the things that we're doing like epilepsy we have reached the stage where we can see how things are generated and synchronized and how it can be stopped okay so many clues have been discovered there where that now is almost ready to hit the clinic where we can now start to help uh them help the patients directly alzheimer's i should have asked about epilepsy first then that that's very exciting yeah and epilepsy and parkinson's and alzheimer's also the key circuitry regarding for example how the motor control is disturbed in parkinson's we're now able to make models around that and so that is one of the things that we're also looking into deploying and alzheimer's disease how different components of cognitive function is disturbed is also now being discovered where we are thinking about ways to deploy that uh although it's a little bit behind the other uh things and we're also looking at different ways of empathy control that allows uh the empathy aspect of the brain disorder can be looked at and so wait a minute that sounds very do you mean the patient's ability to be empathic with other people we're learning how where that is and if it if it gets disrupted we might even be able to kind of help in that area that sounds like a big deal that's why i'm pausing on that absolutely so that is one of the areas that we are also now making uh discoveries and looking into how we can directly help patients and i wanted i didn't want to leave the epilepsy thing because you said some of that is approaching the clinic what is it that we will be doing in the setting of epilepsy that we're not currently doing is it the same technologies but used more precisely or are there new things coming down the pike so right now in epilepsy epilepsy is one of the i mean i don't want to say easy but easier than other diseases you treat right now because epilepsy is much more visible than other diseases and there are some tools doctors can use to measure things with and so with that there are a couple of different treatment options but even there because of the lack of ability to precisely tell what's what the disease onset mechanism is and how it is maintained and how to stop it that's unclear for individual patients and so even though as a epilepsy doctor i have a lot of different tool kits in my bag i just go by try a and see how you feel in three weeks yes bye b we're gonna be able to transform that into the vision that i gave you earlier where we can now say okay this is this patient's status so and i have a much more certain understanding of what to do so toolkit d is a good match so we're going to be able to do matching and will this be a combination of drugs and maybe some surgeries or electrical stimulation absolutely yes really exciting well thank you for listening to the future of everything i'm russ altman if you missed any of this episode listen anytime on demand with the sirius xm app you

2021-02-08 16:41

Show Video

Other news