Scroll Top

How Beacon Biosignals Brings Precision Medicine in Neurology to the Brain

Beacon Biosignals 

Jacob Donoghue, Co-founder and CEO  

Final Transcript  

Harry Glorikian: Hello. I’m Harry Glorikian, and this is The Harry Glorikian Show, where we explore how technology is changing everything we know about healthcare.  

One healthcare trend we talk about a lot here on the show is the rise of precision medicine in neurology. 

Usually that means using information about the particular genes, proteins, and other biomarkers in a patient’s body to choose the therapy that has the best chance of working for them. 

And some of the biggest successes for precision medicine in neurology have come in cancer treatment. 

Using genomic testing, oncologists can often avoid one-size-fits-all approaches like chemotherapy, and instead choose a more targeted therapies such as small-molecule drugs or monoclonal antibodies.  

But there’s one big field of medicine where precision medicine in neurology hasn’t really arrived yet—and that’s neurological and psychiatric disease. 

When it comes to the brain, we just don’t have as many easily measured biomarkers that could help doctors tailor their treatments for problems like epilepsy, Alzheimer’s disease, or depression, or even help researchers develop more effective drugs. 

My guest today, Jacob Donoghue, is the co-founder and CEO of a company that’s trying to change all that. 

It’s called Beacon Biosignals, and it’s focused on making the EEG brain scan into a more reliable and useful data source for diagnosing and treating neurological disease. 

EEG, or electroencephalography, is a non-invasive way to measure electrical activity in the brain, and it’s been a common medical tool for almost 100 years. 

But takes a lot of training for a human doctor to interpret an EEG test for brain correctly.  

It’s slow, it’s expensive, and it’s a bit of a dark art.  

And I bet you know what I’m about to say next. AI to the rescue! 

We’re living through an age when researchers are teaching machine learning algorithms to understand almost every kind of data—and EEG neurology is no exception. 

Donoghue says the goal at Beacon Biosignals is to use computation to get more value out of existing EEG data. 

By peering deeper into the data, he thinks it should be possible to identify subtypes of problems like epilepsy or Alzheimer’s, and help neurologists understand which patients will respond best to which therapies. 

On top of that, better EEG neurological test measurements could also give drug developers and regulators more clinical endpoints to measure when they’re trying to evaluate the safety and efficacy of new drugs for CNS diseases. 

If Beacon’s vision comes true, the precision medicine in neurology revolution might finally start to reach the brain. 

Jacob and I talked back in mid-February, 2023, and I want to play our conversation for you now. 

Harry Glorikian: Hey, Jacob. Welcome to the show. 

Jacob Donoghue: Thanks, Harry, for having me. 

Harry Glorikian: So before we start talking about the new science Beacon Biosignals is working on, I want to start with some real basic stuff here, right? It feels to me like the revolution in genomics and big data over the last 20 years has helped researchers and drug developers make huge strides in understanding the and treating cancer, the immune system, rare genetic disorders and other diseases. But neurologic and psychiatric diseases haven’t really budged. And the treatments we have today for Alzheimer’s or depression or schizophrenia or epilepsy aren’t a heck of a lot better today than what we had two or three decades ago. And some of the medicines that have been developed for Alzheimer’s, for example, have turned out to be very expensive flops. So speaking in broad strokes, I mean, do you agree with that assessment? And if so, why hasn’t neurology, you know, benefited from the same data driven approaches that are helping in other fields? And what do you think have been the main difficulties holding back brain science and making diagnostics and treatment better? 

Jacob Donoghue: That’s a great set of questions. So last decade, you know, clearly was the era of precision oncology connected to precision medicine in neurology. And so really thinking about, you know, matching the right patients based on, you know, molecular and genetic profiles of their disease to the next generation biotechnologies. You know, what is different, you know, in the neurologic and psychiatric diseases is we’re just entering this era where I think we really are on the cusp of better understanding the patient populations and then matching them to the clinical trials and to the treatments that may benefit them. I think, you know, as we’ll talk about in this hour, one of the big difficulties is understanding the heterogeneity of neurologic and psychiatric disease and being able to map that onto outcomes that we can then see movement when we give these new therapies. 

Harry Glorikian: So, yeah, I mean, it’s interesting, right? Because I’m always I just go back in time and I keep thinking to myself, like, right, first it was this big disease, then, oh, no, it’s two. No, it’s three. No, it’s eight. No, you know, and we and we break it off into little chunks, but like we have the tools to sort of start to drill into this. And I’m, I’m thinking like, do we have the tools now to drill into these neurologic areas the way that we would ultimately like to? Or is, or do you believe we’re at the beginning stages of that? 

Jacob Donoghue: Yeah, I think we’re right now at the cusp. So we are starting to appreciate the massive heterogeneity in these neurologic and psychiatric diseases. We’re realizing how important collecting quantitative biomarkers, particular neuro biomarkers about these diseases are. But we really haven’t sort of thought about how we want to measure them in in clinical trials, in late stage, you know, pivotal studies and how those might map to the feelings, function and survival of patients that the FDA cares about. And so, you know, I think this is an exciting time as we’re aggregating data sets to really think about what are the biomarkers that really represent the underlying pathophysiology of disease and how do we how do we make these treatments, you know, make an impact for patients. 

Harry Glorikian: Now, in a minute, we’re going to talk about, you know, how things are starting to change. But, you know, just for everybody on this show, let’s first talk about, you know, electroencephalography or EEG. Right. You know, for those people that are listening, you know, what is the brain function test EEG? What is it? What does it measure in the brain? How is EEG different from other tools like MRI or fMRI? You know, what are its advantages, its limitations, that sort of thing? 

Jacob Donoghue: Yeah. So EEG electroencephalography is the sum of millions of neurons firing electrical activity, recorded non-invasively from the scalp in a person in particular in patients getting, you know, EEG monitoring for diagnosis of epilepsy as a routine part of clinical care, getting a polysomnogram as sleep studies and also and of course, research studies as well. And so it has great time resolution. You can understand the firing activity of neurons over time and can measure really discrete changes in states like encephalopathy when the brain is sick and and in health and in particular the way that brain states change during sleep. So one of the ways that this, you know, clinical grade diagnostic is utilized, as I mentioned, making these diagnoses in epilepsy, making diagnoses in sleep, and where they sort of differ in its sort of resolution. Versus MRI, which allows, you know, high spatial resolution, but nothing over time, you know, a picture of the brain or an fMRI measuring blood flow activity mapped to the brain, which is sort of not able to be done at home. You know, it happens in a magnet and really has a gross sort of correlation potentially to to brain activity. But EEG, which is really, you know, this exciting modality that’s been around for 100 years since the first, you know. What’s changed is we now are gaining the ability to do this more at scale and use machine learning to actually unlock the full depth and breadth of information about brain activity. 

Harry Glorikian: So I know EEG is is, you know, used commonly to diagnose, as you said, epilepsy or sleep disorders. I mean, I have sleep apnea and they like hooked up you know, I was like, can I sleep with all this stuff? I’m hooked up. And they said, Honey, if you have sleep apnea, you’ll be asleep in 30 seconds. So but what makes you believe that it might also be useful for diagnosing or measuring other problems like depression or Alzheimer’s? 

Jacob Donoghue: So I mean, really, at its core, what we’re, the underlying basis of many of these disorders are their genetics. We think of neurophysiology as this bridge between the, you know, molecular and genetic level of disease and the patient reported outcomes and cognitive assessments and sleep diaries. And that, of course, the way we feel and function is governed by brain activity. And so what EEG allows us to do is really track that brain activity over time. 

Harry Glorikian: So, I mean, when I you know, of course, to the untrained eye, right, an EEG looks like just a bunch of squiggles on a page. Right. What kind of training do human neurologists need in order to interpret those squiggles? I mean, I’m thinking of it similar to an EKG, but there’s probably a lot more squiggles in an EEG, if I think about it. Right. It seems that even for an expert, it’d be hard to look at that mess and see anything of clinical significance, especially for conditions like epilepsy that, you know, causes storms of neural activities all over the brain. It seems like making sense of that data just isn’t human, human the ability of a human to actually figure that out. 

Jacob Donoghue: Yeah. So let’s work backwards. So as I mentioned, it is this gold standard for for diagnosing sleep and epilepsy. So actually, in any small subset of time, you know, human experts are, you know, epileptologists, sleep doctors are trained through multiple years of fellowship after their residency, after their medical school to be able to pattern-match, using pattern recognition to understand, you know, this waveform or set of waveforms from a set of electrical leads attached all over the scalp, that that is correlated with a particular syndrome or symptoms of a patient and is indicative of a given disease. And so what’s interesting and different about this moment in time is that the way that these experts spend their precious minutes and hours, you know, reviewing these records is often just to get a gestalt. You know, look at small segments of these, you know, eight hours, 24 hour EEGs to be able to actually make their clinical assessment and use subjective terminology to say things like “The epileptiform spikes were abundant.” And so what’s different in time now is that we actually are starting to have tools that are working as well as experts, as we do at Beacon, algorithms to go and actually find the same types of things that a human might call abnormal and quantify them and not just count them, but look at their derived features and see how they map to subtypes of the disease, how they change with treatment, all the things that you need as a part of an end to end platform to understand which patients might respond to therapies and get them to the right patients. 

Harry Glorikian: And for, you know, anybody that listens to this show like looking at patterns and using machine learning and finding little nuances and, you know, this is where everything is going. And so that finally leads us to the mission here at Beacon Biosignals. And from what I can tell, the mission is to make it easier to gather and analyze EEG readings in order to help identify different subtypes of brain disease and monitor disease progression in the clinical trials participants. And is that a fair summarization? 

Jacob Donoghue: Yeah, I think that’s fair. I think, you know, to restate it, it really is an end to end platform to accelerate drug development for the brain. And I think that starts with understanding your patient populations and leaning on, you know, parts of the Beacon platform like our data store and really being able to characterize what are the waveforms in these different subtypes of diseases, which biomarkers are present in these patients. Again, gold standard, you know, clinical diagnostic level biomarkers that we can anticipate that might be present in one subtype of the disease, like patients with depression and insomnia versus patients without insomnia and depression. And, you know, from there, when you start in this translational research phase, understanding your disease and then being able to have, you know, the playbook and the set of algorithms that you can run to understand these populations, you get to test it on the practice field and then bring those into phase 1, phase 2 clinical trials where you can run our algorithms to be able to understand things like target engagement. Is this therapy making it into the most complex organ in the known universe? And is it modulating activity in the way that you might hope? And then, you know, as we’re moving towards thinking about quantitative neuro biomarkers as real endpoints that really represent the underlying neural pathophysiology, we think about how do we work with the FDA to have quantitative endpoints like epileptiform activity burden that might represent, you know, an endpoint in and of itself, to get away with subjective seizure-diary-like endpoints. So really across the stack, understanding the diseases, understanding their heterogeneity, mapping out these biomarkers in their variability, putting them into play, into clinical trials and seeing new treatments move them so that they’re mapped best to getting these therapies to patients faster. 

Harry Glorikian: So for everybody, the new thing is not the EEG itself, but the ability to analyze it. And actually not a single one, but on large scale. And so, you know, going back to sort of the earlier question about the intractability of like EEG data, why does machine learning offer sort of that new hope of untangling this data? 

Jacob Donoghue: Yeah. So I think in two ways. One, we have really large data sets now that have never been aggregated that we can learn the types of features across diverse patient populations that can then get mapped onto individual patients so we can really see what matters at a scale with precision medicine in neurology – in a way that wasn’t previously possible. And what are all the subtypes of abnormal waveforms in a patient with epilepsy that we can now run inference and detect on a on a particular patient? Secondly, one of the big advantages of having, you know, modern machine learning frameworks for analyzing data is really time, you know, and that matters to our biopharma partners who think about every day of running, you know, multi-million dollars for their trials. And with one example, you know, you know, manual labeling of sleep staging in in a drug, for example, for insomnia might take something like six months to get the end points calculated. Did this drug improve sleep metrics in this patient population? With modern tools, with what we’re doing now, we can rerun, you know, ten, twenty thousand polysomnograms with state of the art sleep staging and it takes a couple of hours. So it could be months on patent and insights that you get, obviously, drugs through development faster, but most importantly, treatments to patients, you know, months earlier. 

Harry Glorikian: And I would assume that while you’re looking at that, you’re identifying other features in there that might be relevant to, you know, the next drug, let’s say, that’s coming forward. So let’s go to the founding story of Beacon Biosignals, you know. How did you first get interested in the overlap of neurophysiology and machine learning? You know, you’ve got four founders at the company accounting yourself, which is a lot, right? It’s usually one or one or two. And it’s you, Jarrett Revels, Brandon Westover and Sydney Cash. I mean, how did you guys come together and what gives you guys the unique experience or insights to put this together? 

Jacob Donoghue: Yeah, well, you know, sort of locked in from the start as the son of a pediatric neurologist and a neuroscientist, I think sort of this attention and interest in brain activity began early. But really, you know, getting to work with Syd Cash and his lab at Mass General Hospital, coming out of Brown, really understanding, you know, putting electrodes in patients with epilepsy or going for surgery, really seeing just how barbaric, you know, our treatment methodologies really are. We don’t have, you know, modern biotechnologies for some of these disorders and we have to literally resect parts of the hippocampus or the cortex of the brain. And so, you know, as I started my MD PhD at Harvard and my PhD work at MIT, getting to work with Emory Brown and Earl Miller, working with, you know, nonhuman primates, studying drug effects on the brain, looking at how anesthetics changed global network activity. And what really was remarkable is just how rich this information is that you could predict even by eye, seeing the waveforms change at the dose of which particular mechanism of the drug you gave. And so, you know, and obviously machine learning tools allow you to do a much better job to be more quantitative about understanding that. And so I think that, you know, in short, that transition, when you go back to the clinic and start taking care of patients, one of the beautiful things about being a medical student is getting to see, you know, what does precision oncology look like in terms of precision medicine in neurology, you know, in 2022. 

Jacob Donoghue: And you see patient, matching patients to the right therapies and what is, you know, quantitative at home monitoring look like for, you know, cardiology where we’re sending patients home, you know, with, no longer with Holter monitors, with things like patches. And then arriving at neurology and psychiatry and really feeling frustrated that, you know, seeing patients, as we talked about before, experts reviewing EEGs with, you know, hundreds or thousands of spikes in really high epileptiform activity burden. And the experts just saying, well, the spikes are abundant and, you know, and what is a new treatment doing to that? Well, it’s still abundant. But was that 20% reduction meaningful for that two year old with autism? You know, maybe that finding. You’ll see in a few years that they actually hit their milestones earlier. So this lack of tools and this sort of ignorance of the way that people are thinking about developing new treatments in oncology and the way we take care of, you know, patients where we’re understanding the disease at its, you know, most precise level was really transforming medicine, brought this opportunity where we came together with myself and Sydney and Brandon Westover, whose lab had done a lot of machine learning on EEG at scale. And then and then Jared Revels, the CTO and co-founder, you know, orchestrating, you know, these large scale machine learning models and setting up data infrastructures to be able to really understand these populations at scale really sort of comes together in a perfect way. 

Harry Glorikian: Yeah, it seems like there’s more and more coming across my desk in this space. Like, you know, I interviewed Rune on their Parkinson’s, right? I talked to Kernel, which is making sort of an at home helmet, right? Trying to just talk to different people to understand like, okay, what’s what seems like it’s working, what doesn’t seem like it’s working, What’s your approach. And you can see how you can help somebody make therapy more precise for, like, we keep overusing this word, but, you know, at least some guideposts. Does it look like it’s working or not? And in Rune’s case. But can you talk a little bit about the you know, what is the special sauce at Beacon? Right. Is it you know what is it a novel algorithm you’ve come up with to do, you know, machine learning on these particular, you know, on the EEG data or trying to figure out like what makes you different versus somebody else coming into the space? 

Jacob Donoghue: So really, it’s our platform, the ability to ask questions about diseases of the brain and have answers, you know, in hours, weeks, not months or years. And so built on this foundation of the world’s largest clinical EEG data sets. So real world patient data with EEGs linked to medical records and ICD codes and medications. So we can really understand why certain biomarkers are in certain subtypes of patients and then having that foundation laid so that you can actually run the same type of analytics in clinical trials and see those endpoints get moved with a new treatment. So I think that software infrastructure allows us to again answer those questions. 

Harry Glorikian: So I guess that makes me want to ask why clinical trial and not the diagnostic path? 

Jacob Donoghue: Yeah. So early on, there’s obviously a huge unmet need on both sides here. And the main goal for Beacon is accelerating precision medicine in neurology therapies to help patients. And what we find here is that we are, you know, deeply embedded in the science and really understanding how individual biomarkers can change and how that might be related to describing these diseases. And so we kind of have to set the standard as we think about, you know, precision diagnostics in terms of precision medicine in neurology, companion diagnostics and thinking about how that clinical landscape is going to look in 5 to 10 years when we anticipate monitoring, you know, 500,000, a million patients getting routine brain monitoring as part of their clinical care. But the alignment with biopharma allows us to really focus in on the science and deliver immediate value right away. So we’re being rewarded for that, for that alignment early on.  

[musical interlude] 

Harry Glorikian: Let’s pause the conversation for a minute to talk about one small but important thing you can do, to help keep the podcast going. And that’s leave a rating and a review for the show on Apple Podcasts.  

All you have to do is open the Apple Podcasts app on your smartphone, search for The Harry Glorikian Show, and scroll down to the Ratings & Reviews section. Tap the stars to rate the show, and then tap the link that says Write a Review to leave your comments.  

It’ll only take a minute, but you’ll be doing a lot to help other listeners discover the show. 

And one more thing. If you the interviews we do here on the show I know you’ll my new book, The Future You: How Artificial Intelligence Can Help You Get Healthier, Stress Less, and Live Longer.  

It’s a friendly and accessible tour of all the ways today’s information technologies are helping us diagnose diseases faster, treat them more precisely, and create personalized diet and exercise programs to prevent them in the first place. 

The book is now available in print and ebook formats. Just go to Amazon or Barnes & Noble and search for The Future You by Harry Glorikian. 

And now, back to the show. 

[musical interlude] 

Harry Glorikian: So. What are you learning so far about whether diseases like epilepsy or Alzheimer’s have unique EEG signatures. I mean, what about schizophrenia, depression or other sleep disorders?  

Jacob Donoghue: So you start with epilepsy. Epilepsy affects 65 million people worldwide. Half a million kids in the US. And so, again, a really heterogeneous disorder. There are patients with temporal lobe epilepsy which may have, you know, rare seizures and there may be, you know, children with developmental epileptic encephalopathies that have, you know, unbelievably high seizure and spike burden. So that heterogeneity is something that, you know, we really want to characterize here at Beacon. You know, for patients and for our biopharma partners making precision therapies connected to precision medicine in neurology. So one of the things we’re finding is that there’s actually, unsurprisingly, a huge gap on the underlying epileptiform activity burden. You know, the EEG abnormalities and what’s being reported in seizure diaries. And this maybe isn’t so surprising. You know, some patients, in particular children with these disorders aren’t able to report their seizures or have seizures overnight when no caregiver is reporting them in a diary. And even if, you know, some of these patients have absence seizures, where sort of a brief staring spell might have might actually be a clinical seizure, where if you had an EEG during that time, you’d see frank abnormalities. And so starting to collect these data sets and start linking what are we measuring as our patient reported outcomes data and seeing just how that that delta between what’s actually going on in the brain is helping us understand the real signals that are there that might be treatable, and what we’re really missing by looking at the at the surface of subjective reporting. And so what we find some of the most exciting results that are that are coming out of Beacon, we’re seeing patients with differing levels of epileptiform activity burdens. So some kids with high spikes and some kids with low spikes with two different with the same, you know, two different mutations of the same gene. And we’re finding links that the kids with high spikes may have worse outcomes, missing more milestones. And so really seeing that this electrographic signature of the disease is linking to the genetics, all the way to the patient outcomes data. And so I think that epilepsy provides this great starting point for us where the signals are known to clinicians as as being part of the diagnostic journey of understanding the disease, but never with this precision of linking such, you know, fine-grained molecular and genetic anomalies to the outcomes of the patients. And so we think that starting here in epilepsy allows us to then map out, as you called, out, psychiatric disorders such as depression and schizophrenia. 

Harry Glorikian: So it’s not just necessarily the analysis of the EEG, EEG data, but it’s also all the other data that you’re bringing in that’s sort of helping paint the a bigger picture. 

Jacob Donoghue: Absolutely. So again, our platform serves as this multi-modal hub centered around, you know, time series, neurophysiologic data EEG and polysomnogram, but really be able to tie in other types of metrics such as voice or actigraphy to be able to really map out and link to the outcomes data. 

Harry Glorikian: So here’s the, you know, being the venture guy, right? The multi-million dollar question or maybe I should say multi-billion dollar question is how might insights from these EEG signatures speed up drug development or lead to the discovery of, say, a more precise treatment for a neurological disease? Can you give an example of how you think this is going to play out over time? 

Jacob Donoghue: And to bridge to your last question, I think one one exact instantiation we think is in Alzheimer’s disease, where we’re having increasing number of work looking at brain activity in patients with different forms of of Alzheimer’s. And so there what we’re finding and replicating, replicating at a larger scale, some of the early results that were published over the past five years, is that patients with Alzheimer’s disease, some of them when you get an EEG, even though they’ve never reported a clinical seizure, they actually have subclinical epileptic activity, epileptiform activity. So you look on the EEG and they have sharp waves by eye. You can see them. And we’re seeing, you know, reported in between 20 and 40% of patients. And that’s sort of what we’re seeing. And the question is, so what do those matter? Well, what’s interesting is you when you look at those patients over time, you know what’s been published by groups like Vossel and others, that that those are the patients that actually have worse outcomes faster. So their cognitive, they precipitously decline in their in their cognitive abilities. And so what does that mean for, you know, where’s the billion dollar idea for drug development as we’re seeing? Well, if you know that there’s a 20 to 40% of patients with Alzheimer’s disease that are declining significantly faster, five times faster than the other group, well, maybe they shouldn’t be in that clinical trial. Maybe they are masking the effect of great beta amyloid antibody therapy. Or alternatively, maybe you have a treatment that actually targets those patients better and maybe you do not use six months or 12 months cognitive test as an endpoint, you measure EEG and look for those spikes to get reduced and you do a two week long phase two clinical trial. So that mentality of thinking, you know, just in Alzheimer’s, we sort of can can apply across the board in neurologic and psychiatric disease. 

Harry Glorikian: So do you think that the EEG data could be a substitute or maybe a supplement to other kinds of endpoints in clinical trials? 

Jacob Donoghue: Absolutely. And we’re I think, in fact, a surrogate endpoint, to be specific. We think that there are, you know, pharmacodynamic response biomarker elements to the way that something even like spike burden and epileptiform activity burden, as we discussed, could be modulated by a novel therapy. 

Harry Glorikian: So, are the makers of the therapeutics or the FDA accustomed to using EEG data for as a clinical endpoint. I mean, when they’re evaluating, let’s say, a pivotal phase 3 trial. And if they’re not, how do you get them used to this? 

Jacob Donoghue: Yeah, I think you know the answer to the latter part. So I think our biopharma partners are very excited. They understand that this really the neural activity is underlying some of the core elements of the syndromes and the diseases that they’re studying. Of course, what matters at the end of the day when making therapies is do you hit your endpoints? And so I think we do have a, you know, a tall order to to fill here of really validating and qualifying these biomarkers to show their links to the disease beyond what is clinically known and accepted, you know, EEG, seizures, EEG spikes to be actually really qualifying them as endpoints in clinical trials. And so, you know, we have a really targeted regulatory strategy, thinking about how to work with the FDA and programs like the qualified biomarker program to move forward, epileptiform activity burden, eye guided sleep metrics to become endpoints that our biopharma partners can safely reach for for their pivotal studies, not just phase 1 and 2, where they can de-risk business decisions, but really, you know, at the end of the road, get get a drug approved because it’s, you know, transforming, you know, in a disease modifying way, you know, brain activity. 

Harry Glorikian: Yeah. I mean, it’s you know, I think it’s important all the way around. Right? You want to fail fast If you’re going to fail, you don’t want to fail at the end. That’s super expensive failure. But if you have something where it’s really showing that it’s working right, you want to be able to show that data to regulatory authorities, physicians, etcetera, that we’re actually moving the needle. So I think it’s valuable all the way around. But I mean, we all know that machine learning algorithms learn way better from lots and lots of data and typically label and annotated by, you know, humans. I think on the website it says you guys have data from 50,000 people with 3.8 million annotations, right? I’m sure that number, you’ve got to build a clicker because I’m sure it’s growing over time. But how’d you guys build up the data or, you know, I don’t know where you got it from, but who labeled it? Give me some stats. 

Jacob Donoghue: So again, so any any machine learning, you know, really is only going to be as good as the training data that it was built upon. And for us, I think that’s one of the key differentiators. The world’s largest is only one element. It’s the fact that we’ve spent so much data engineering efforts to to build out and curate this data. And so through health system partnerships, academic partnerships and biopharma partnerships, we’ve continued to aggregate, you know, data sets that we can link back to the, you know, medical records for patients and getting annotations, which can be done two ways. One is through routine clinical care. So that’s serves a large percentage of our of our data. But also we have an entire Web based EEG and PSG labeling platform that we can use, whether for our biopharma partners, when we need human in the loop to adjudicate a subset of the recordings or for ourselves. We’re studying a new disorder like epileptiform spikes in Alzheimer’s disease, we want to make sure our algorithms are really generalizing well to new forms that might not have been in the training sets. So we can add new labels as we go. And, you know, pretty much every day, you know, more and more labels are getting added to the data sets. 

Harry Glorikian: So how many of these people have a CNS disease and how many are healthy because you need to have healthy people in there, right, to be able to create a baseline. 

Jacob Donoghue: Yeah, that’s great point. So, you know, through through research studies and through, you know, ordering of EEGs and polysomnogram that are found to be otherwise normal patients that have no, you know, comorbidities or taking a medication that might affect brain activity. We actually have close to 5,000 patients that represent healthy, normal neurotypical brain development. And that is, you know, unique. The way we can use that is in a few different ways. Again, first off, you know, this patient age ranges from zero to hours old to 100 years old. So we need to do, as you called out, benchmarking. What does normal brain activity look like? We’ve done this for EKGs and we know sort of the metrics by which normal variability can be represented, but nobody’s really had this framework for brain activity. And so we can do things like measure slow wave power in non-REM to sleep and see what that does over, you know, the development of a typical child’s growth. And why that’s really critical for understanding disease is you want to basically benchmark and Z score any particular feature from the healthy controls. And so the way we use that with our pharma partners both is to one, understand just how abnormal a particular disorder might be and then really understand the disease modifying properties. One example might be in kids with some of these developmental and epileptic encephalopathies and really rare seizure disorders. We look at the background and there’s large slow waves. So big oscillating, you know, delta waves as they’re called. But, you know, actually young babies can have some of these, too, before the fontanelles of the skulls close. And how do you really differentiate? Is this normal or abnormal? Well, you really need to understand the age matched control. And so we can actually do that, you know, adjustment and understand exactly how abnormal a particular feature is and then set that, you know, pharma partner up for success when they give that new therapy and see does that feature move back towards normal, helps understand the mechanism of the drug scientifically and potentially even for some of our other partners, really at a commercial commercial level, you know, we can create the first therapy of this kind that causes normal, healthy sleep, normal, healthy, healthy, awake brain activity. 

Harry Glorikian: You know, almost like, although you’d really want to follow the same person over time, is how things change in their brain waves over time. Right. And almost want to feel like I want to do it myself just to see what’s going on up there. 

Jacob Donoghue: Well, and that front, I mean, you know, many of the patients that, you know, that get an EEG often get multiple EEGs over their lifetime. And so we are able often to see how an EEG at age one predicts outcomes at age two, three, four and beyond. 

Harry Glorikian: That’s cool. That time series part is is very cool. I mean, I have sleep apnea, so I’ve, you know, sometimes I wish there was a better way to categorize it. Right. Right now it’s like you have sleep apnea. Here’s the CPAP. Go home. But there’s there’s way more to it that just isn’t necessarily being, you know, understood or dug into. So how long do you think it might be before an EEG study is part of standard of care for patients with suspected, you know, neurophysiological problems, whether that means sleep disorder, epilepsy, neurodegenerative problem or psychiatric problem? I mean, and will it eventually become standard operating procedure to run this data through machine learning analysis? 

Jacob Donoghue: Yeah. So it’s important to remember that, you know, EEGs are part of the standard of care for epilepsy and sleep medicine. The problem is, is that they are expensive, hard to scale and don’t really reach a lot of the patients. And so therefore, oftentimes many of these diagnoses can be made clinically as opposed to having the quantitative data to really make the definitive diagnosis or understand the impact of a novel treatment. So I think, you know, you can imagine a world one that Beacon hopes to power where, you know, every pediatric neurologist office or maybe even a pediatricians office is able to order EEGs to understand why a kid is missing some of its motor milestones, to really see what’s going on in brain activity related to some of these disorders. Same with sleep. You know, I think, you know, because of the way the reimbursement models go at home, brain monitoring really only gets part of assessment for sleep apnea. So as a result, there’s no financial incentive, you know, or care alignment from payer perspective and provider perspective to reimburse a sleep study for somebody that has depression and insomnia, depression and hypersomnia or schizophrenia. And so I think what we hope is that actually that we will be able to bridge out of just epilepsy and sleep apnea when we start to prove with our biopharma partners just how rich and linked brain wave activity is to some of these other disorders and how it really can help predict outcomes, help predict treatment, efficacy, that we easily envision a future where this is a routine part of care, much at a much bigger scale and outside the area as it currently is. 

Harry Glorikian: And you’d love the price of doing the actual test to come down and be easier and almost have your platform built in. 

Jacob Donoghue: Absolutely.  

Harry Glorikian: All right. So is that what you guys see for the future of the company?  

Jacob Donoghue: Yeah. So I think there’s a virtuous cycle here. We can understand the science and which waveforms are most strongly linked to the disease phenotypes. Which ones can move with treatment to help get novel precision therapies connected to precision medicine in neurology brought to patients faster really have an impact on patients with huge unmet need. And then really proving out that in the clinical context you need this brain monitoring to be able to understand which patients get which medications. Again, this precision oncology  framework connected to precision medicine in neurology just mapped to our world. And so we see a future easily where, you know, a novel antisense oligonucleotide is given to a child with a rare epilepsy like SYNGAP-1 related disorders. And, you know, we give the ASO and we don’t say, come back in six months with your seizure diary. Well, you were given routine at home, EEGs, you know, multiple times per week. And we saw the epileptiform activity burden start to creep up at, you know, three weeks after or three months after. And that’s when you come back in for that second dose, not something arbitrary that’s based, you know, at a wider patient population. So really targeted precision therapies connected to precision medicine in neurology individualized to patients that we can do in a future where we have these biomarkers validated and enabled that scale. 

Harry Glorikian: Well, this has been great. I mean, I’m glad we we got this on the calendar and we made it happen. I mean, it’s funny because I do have a conversations with my neurologist friends and I’m like, Dude, you guys are still in the dark ages, relatively speaking, and they’re always interested in like, what’s coming next and who am I talking to and so forth, because there’s just not a lot right? And they’re waiting for all the papers to come out and prove it before they can actually do something with it in their practice. I mean, by the time it trickles down, it takes forever. 

Jacob Donoghue: Yeah. I mean, I think what’s really exciting about this time is that we actually have some of the biotechnologies on the cusp of being delivered. And, you know, starting with Spinraza for SMA, we have targeted therapies coming in, some already approved for some of these disorders. We just need the bridge to better understanding their efficacy so that we can really make sure that the ones that are having an effect on the brain get approved. And and those neurologists are able to prescribe those drugs to the right patients. 

Harry Glorikian: Right. And then that it’s working afterwards that you can follow up and show that it’s, you know, the you can adjudicate whether it’s really happening or not. 

Jacob Donoghue: Absolutely. 

Harry Glorikian: So great having you on the show. I’m looking forward to staying in touch and hearing how things are progressing. 

Jacob Donoghue: Thanks so much for having me. 

Harry Glorikian: That’s it for this week’s episode.  

You can find a full transcript of this episode as well as the full archive of episodes of The Harry Glorikian Show and MoneyBall Medicine at our website.  

Just go to and click on the tab Podcasts. 

I’d to thank our listeners for boosting The Harry Glorikian Show into the top two and a half percent of global podcasts. 

To make sure you’ll never miss an episode, just open Apple Podcasts or your favorite podcast player and hit follow or subscribe.  

Don’t forget to leave us a rating and review on Apple Podcasts.  

And we always love to hear from listeners on Twitter, where you can find me at hglorikian. 

Thanks for listening, stay healthy, and be sure to tune in two weeks from now for our next interview. 

FAQs about Precision Medicine in Neurology

What is precision medicine in neurology?

Precision medicine in neurology refers to the application of personalized approaches in the diagnosis, treatment, and management of neurological disorders. It aims to tailor medical care to individual patients based on their specific characteristics, such as genetic makeup, environmental factors, lifestyle, and biomarkers. By utilizing advanced technologies and analyzing large sets of data, precision medicine enables healthcare professionals to provide targeted therapies and interventions, leading to more effective and efficient treatment outcomes for patients with neurological conditions.

Precision medicine in neurology involves several key aspects:

  1. Genetic Profiling: Genetic testing plays a significant role in precision medicine. By analyzing a patient’s genetic information, including DNA sequencing, researchers can identify specific genetic variations or mutations associated with neurological disorders. This information helps in diagnosing certain conditions, predicting disease progression, and determining treatment responses.
  2. Biomarkers and Diagnostics: Precision medicine utilizes biomarkers, which are measurable indicators in the body that can reflect normal or abnormal biological processes. Biomarkers can help diagnose neurological disorders, monitor disease progression, and assess treatment response. Neuroimaging techniques, cerebrospinal fluid analysis, blood tests, and other diagnostic tools help identify relevant biomarkers.
  3. Targeted Therapies: Precision medicine aims to develop targeted therapies that are tailored to an individual’s specific disease characteristics. For example, certain genetic mutations may render a patient more responsive to a particular medication or treatment approach. By identifying these genetic markers or other relevant biomarkers, physicians can select the most appropriate treatment options, maximizing therapeutic benefits and minimizing adverse effects.
  4. Disease Subtyping: Many neurological disorders, such as Alzheimer’s disease, Parkinson’s disease, and multiple sclerosis, have heterogenous presentations and underlying causes. Precision medicine aims to identify disease subtypes based on various factors, including genetic variations, clinical features, imaging findings, and biomarkers. This subtype classification helps in selecting the most appropriate treatment strategies and predicting disease progression.
  5. Data Integration and Analysis: Precision medicine relies on advanced technologies for data integration, including electronic health records, genomic data, imaging data, and other relevant clinical information. Machine learning and artificial intelligence algorithms analyze these large datasets to identify patterns, make accurate predictions, and develop personalized treatment plans.

Overall, precision medicine in neurology aims to improve patient outcomes by providing targeted, individualized care based on the unique characteristics of each patient’s neurological condition. It has the potential to revolutionize the field of neurology by optimizing treatment strategies, accelerating drug development, and advancing our understanding of neurological disorders.

What is an EEG brain scan?

An EEG (electroencephalogram) is a diagnostic test that measures the electrical activity of the brain. It records the brain’s electrical impulses through small metal electrodes placed on the scalp. An EEG brain scan is a non-invasive procedure that helps in evaluating and diagnosing various neurological conditions by analyzing the patterns and characteristics of the brain’s electrical activity.

During an EEG, the electrodes are attached to specific locations on the scalp using a sticky gel or adhesive. These electrodes detect the electrical signals generated by the neurons in the brain. The signals are then amplified and recorded by the EEG machine, which displays them as a series of waveforms on a computer screen or printed paper.

The EEG provides valuable information about the brain’s electrical activity and can help diagnose or monitor several conditions, including:

  1. Epilepsy: EEG is commonly used in the diagnosis and management of epilepsy. It can detect abnormal electrical discharges or spikes in the brain, known as epileptiform activity, which are characteristic of epilepsy.
  2. Sleep Disorders: EEG is also used to assess sleep disorders, such as sleep apnea, narcolepsy, and insomnia. It helps identify patterns of brain activity during different stages of sleep, including rapid eye movement (REM) sleep and non-rapid eye movement (NREM) sleep.
  3. Brain Disorders: EEG can provide information about various brain disorders, such as brain tumors, brain infections, strokes, and degenerative disorders like Alzheimer’s disease.
  4. Brain Injury: In cases of traumatic brain injury (TBI) or coma, EEG can help assess the severity of the injury, monitor brain function, and predict outcomes.
  5. Evaluation of Consciousness: EEG can assist in assessing the level of consciousness in individuals with conditions like coma or disorders of consciousness.

The interpretation of an EEG requires expertise, as the patterns and abnormalities can vary depending on the specific condition being evaluated. Neurologists or neurophysiologists typically analyze the recorded EEG data to make a diagnosis and guide further treatment decisions.

In summary, an EEG brain scan is a non-invasive test that measures the electrical activity of the brain. It is a valuable tool in the diagnosis and management of various neurological conditions, particularly epilepsy, sleep disorders, brain disorders, and brain injuries.