Data Science and AI Applications in Managing Subarachnoid Hemorrhage Patients

The critical care management of spontaneous subarachnoid hemorrhage (SAH) is similar to that of other acute brain injuries, with the addition of detecting and treating delayed cerebral ischemia. Recent trials are influencing practice and providing guidance for standardizing management.

In this episode, Kait Nevel, MD speaks with Soojin Park, MD, FAHA, FNCS, author of the article "Emergent Management of Spontaneous Subarachnoid Hemorrhage," in the Continuum June 2024 Neurocritical Care issue.

Dr. Nevel is a Continuum® Audio interviewer and a neurologist and neuro-oncologist at Indiana University School of Medicine in Indianapolis, Indiana.

Dr. Park is an associate professor of neurology (in biomedical informatics) at Vagelos College of Physicians and Surgeons, Columbia University in New York, New York and medical director of critical care data science and artificial intelligence at NewYork-Presbyterian Hospital in New York, New York.

Additional Resources

Read the article: Emergent Management of Spontaneous Subarachnoid Hemorrhage

Subscribe to Continuum: shop.lww.com/Continuum

Earn CME (available only to AAN members): continpub.com/AudioCME

Continuum® Aloud (verbatim audio-book style recordings of articles available only to Continuum® subscribers): continpub.com/Aloud

More about the American Academy of Neurology: aan.com

Social Media

facebook.com/continuumcme

@ContinuumAAN

Host: @IUneurodocmom

Guest: @soojin_soojin

Full episode transcript

Dr Jones: This is Dr Lyell Jones, Editor-in-Chief of Continuum, the premier topic-based neurology clinical review and CME journal from the American Academy of Neurology. Thank you for joining us on Continuum Audio, a companion podcast to the journal. Continuum Audio features conversations with the guest editors and authors of Continuum, who are the leading experts in their fields. Subscribers to the Continuum journal can read the full article or listen to verbatim recordings of the article by visiting the link in the show notes. Subscribers also have access to exclusive audio content not featured on the podcast. As an ad-free journal entirely supported by subscriptions, if you're not already a subscriber, we encourage you to become one. For more information on subscribing, please visit the link in the show notes. AAN members, stay tuned after the episode to hear how you can get CME for listening.

Dr Nevel: This is Dr Kait Nevel. Today, I'm interviewing Dr Soojin Park about her article on emergent management of spontaneous subarachnoid hemorrhage, which is part of the June 2024 Continuum issue on neurocritical care. Welcome to the podcast. It's so great to be talking to you today.

Dr Park: Thank you so much, Kait. Nice to be here.

Dr Nevel: Before we get started, could you introduce yourself for the audience?

Dr Park: Sure. So, I am an Associate Professor of Neurology - also in Biomedical Informatics - at Columbia University here in New York City. I trained in vascular neurology and neurocritical care.

Dr Nevel: Great. And so, I always like to ask at the beginning of these interviews, you know, if we could take away one thing from your article — and this is specifically (I'll direct this) towards the neurologists out there that are covering inpatient consults and ER consults — and so, for our clinical neurologists listening out there, what is the most important thing that you think that they should take away from your article?

Dr Park: So, I guess the most important thing for the general neurologists out there is that it may have been a while since they were aware of some updates that have occurred. There are some recent trials that are influencing practice and will potentially influence practice in the next few years that readers should really know about, and it provides a little bit stronger guidance to drive more standardized management. There have been two recent guidelines published this year. But there remain several gray areas for management where you need to be a bit more nuanced, and so I'm hoping the article gives the readers a framework to deliver more expert care.

Dr Nevel: Yeah, and I really, of course, always urge the listeners to go back and read the article and reference the article, because I do think that you do that really nicely and are clear when there are things where there's more higher-level, evidence-based reasons for things and where there's, kind of, just more expertise and guidelines on certain things. So, could you tell the listeners a little bit more about yourself, what interests you about subarachnoid hemorrhage specifically, and how you approach that interest and clinical background in writing this article?

Dr Park: So, I mentioned that I trained in both vascular neurology and neurocritical care back when many people used to do that. As a result, I've trained or practiced in four different academic medical centers who have specialized neurointensive care units. And the patients with subarachnoid hemorrhage tend to have a substantial ICU length of stay, and the neurointensive care that we provide can have a very large impact on patient outcome. And what I saw, though (practicing across four different centers), was that the management of patients with subarach can be quite variable across institutions and across patients within institutions, and it's reflective of a couple of things. One, there's, like, complexity in detecting ischemia, even when your patient is a captive audience in their ICU room. Second, there's many clinical mimics that occur (the patients with subarachnoid hemorrhage, they have a risk for), such as hydrocephalus, seizure, and things like delirium. And then, finally, there's limitations in the technology that we even have available in terms of monitoring these patients. But, for me, it was this complexity and the variability of management that kind of posed an opportunity, and it really sparked my curiosity early on and has sustained me. So, I'm particularly interested in the role that, kind of, the complex analysis of existing monitoring technologies can play to improve outcome for patients with subarachnoid hemorrhage, and that's where the marriage of both being a neurointensive care physician and a biomedical informatics person comes in.

Dr Nevel: Yeah. That's really interesting, and I could see that, because I always felt, even during my training, that some of the management and, you know, what diagnostics were even ordered to follow patients throughout the ICU was expertise based and seemed to vary without a lot of really solid, again, high-level studies, guiding what was done. So how do you marry the bioinformatics with your interest in SAH?

Dr Park: Right. So, I have two grants on - basically, I guess you would say AI, but really data science - on how we can manage patients with bleeds, specifically ICH and subarachnoid hemorrhage and hydrocephalus. So, we use data that comes from the monitors and we process that in a multimodal fashion and apply signal processing and machine learning and we build predictive analytic tools. So, I'm very interested in this pipeline of developing clinical decision support (information that we don't really have), and we're trying to glean from all the data and turn it into information that clinicians might use. The problem in subarachnoid hemorrhage patients is that a lot of what we're looking for is subclinical - so, it's not quite obvious, either because you can't possibly be in the room to be constantly monitoring for it (and, currently, the best monitor is the human, is examination), but, specifically in patients who have disordered consciousness, even the examination can be somewhat limited, and that's where we rely upon some of our neuromonitors. So, my interest has come in taking those multimodal monitors - but even nonneurologic monitors (stuff about your physiology, like your heart rate and blood pressure, et cetera) - and able to find signals that might tell us that a patient is getting into a dangerous zone. So, that's what my research portfolio has been 100% about - it's about subarachnoid hemorrhage patients and trying to optimize management, both for prevention and intervening in a timely fashion.

Dr Nevel: Wow. That's really interesting and would be so wonderful, it sounds like, for this patient population, if, you know, something was able to be identified that you could easily monitor to kind of predict or catch things early. So, kind of segueing from that, what do you think are the most — and you outline these nicely in your article, and I'm going to reference the listeners to, I believe it's the first table (table 5-1) - but what are, just like in general, the most important initial steps a clinician should take when managing somebody with an aneurysmal subarachnoid hemorrhage?

Dr Park: So, I think it's sort of along the timeline. So, at the time of presentation of a patient with subarachnoid hemorrhage, the focus you should have should be really on differentiating the etiology of the subarachnoid hemorrhage. At the same time, if the patient has any coagulopathies, you should manage that coagulopathy reversal, blood pressure management, and then detection and management and treatment of hydrocephalus. That's first and foremost. But then there is a longer timeline of neurocritical care management, and that's really centered on prevention, detection, and treatment of delayed cerebral ischemia, and that can occur anytime from onset of subarachnoid hemorrhage to two to three weeks out. And then that period of neurocritical care is made challenging because you have early brain injury (which is the period of seventy-two hours after onset), cerebral edema, and then, like we talked about, disordered consciousness. This kind of knowing how to augment your management strategies with monitoring or imaging is really key.

Dr Nevel: Yeah. And you, you know, spend some time in your article really going through delayed cerebral ischemia really nicely. And I would love to hear your take on what is the most challenging aspects of delayed cerebral ischemia in both, you know, diagnosis and management - and you alluded to it a little bit earlier, I think, with some of your research, but I would love to hear you talk about that.

Dr Park: Yeah. And actually, this is probably one of - if there was a controversial area in this topic, it would be about this - because there does not seem to be one best way to operationalize how you either survey for, or monitor for, delayed cerebral ischemia. There has been, historically, a merging of these definitions of vasospasm and delayed cerebral ischemia, which are not the same thing. And so, if you were to draw a Venn diagram, not all patients who have cerebral vasospasm end up having symptomatic or delayed cerebral ischemia, and not all patients who have delayed cerebral ischemia have any discernable vasospasm - and, so, to use the terms interchangeably leads to a little bit of confusion. I mentioned the clinical mimics - you know, the causes of which are myriad (could be delirium, or hydrocephalus, or early brain injury) - and so that also poses another challenge. And, so, what I always say is that delayed cerebral ischemia, sometimes - when you're thinking about it in the context of subarachnoid hemorrhage - is sometimes a retrospective diagnosis. And it really kind of came from a really earnest attempt to standardize what the community is talking about, so that we can better understand how to define (if you understand how to define it better, then you can tailor treatments, study treatments, you're talking about the same disease) - but we're still not there, and I think that's where a lot of the controversy or confusion comes from. My personal approach is really to focus on the symptomatology, so, if a patient has vasospasm - whether that is, you know, screened for with a transcranial Doppler (if your institution does use transcranial Dopplers, it might be a nice screening tool) - but the fact of the matter is that not all patients can get a transcranial Doppler every single day. You know, most of the institutions that I have worked in offer that technology Monday through Friday and not on holidays, not on weekends, and so you can't fully rely upon something like that. The advantage of it is that it has pretty high sensitivity but it does have a lower specificity (so it overcalls vasospasms), so to treat just based on a TCD would probably be erroneous. Not all people agree, but I think that's the majority of the sentiment - is that you should then be triggered to go look for confirmation with some neuroimaging and really potentially wait for symptoms so that it might be a trigger to optimize the patient in terms of volume and blood pressure, but not necessarily to treat. So, yeah, operationalizing that workflow of how do you trigger, you know, confirmatory neuroimaging, what type of neuroimaging you should then choose? This is where the variability exists. But, in general, I focus on symptomatology. The extra challenge comes in the patients who have disordered consciousness. And so, at an institution like mine, we do rely upon invasive neuromonitoring, and that's now called for in the guidelines as well.

Dr Nevel: And I imagine these are high-intensity situations where also I would suspect decisions, you know, need to be made quickly on some of these things that you're talking about, too.

Dr Park: That's right.

Dr Nevel: What do you think is a misconception - or maybe (I hate to call it a mistake, but for lack of a better term) like an easy mistake that one can make - when treating patients with aneurysmal subarachnoid hemorrhage?

Dr Park: Hmm, an easy mistake. I guess, you know, time is brain, so it's an opportunity to miss ischemia - or actually attribute everything to ischemia and ignore the possibility for things like seizure (so nonconvulsive seizures), a resurgence of more of a delayed hydrocephalus - and so, I think it's important as you're managing a patient not to get kind of pigeonholed into looking for one particular thing (only looking for delayed cerebral ischemia), but being really vigilant that there could be lots of different reasons for a neurological change of a patient. And so, timely monitoring - kind of figuring out the etiology of a change in neurological status - is really important. And then, also, on the flip side of that, is we're really good at being aggressive in both inducing hypertension or managing a patient (trying to prevent ischemia), we're not that great about starting to pull back - and so I think being vigilant about opportunities to reassess your patient's risk for ongoing ischemia and deciding when that period of risk is over and starting to peel back on therapies, because these patients are also at risk for the down sides of inducing hypertension, which is PRES - and we have seen that in patients, and, you know, the phenotype of that will look very much like ischemia.

Dr Nevel: Yeah, it's complicated because you're taking care of patients with often impaired consciousness who have a lot of symptoms that could represent many different diagnoses that you would treat very differently, so I could see that that might be easy to do to kind of fall into the mindset of thinking that it's definitely one thing without fully evaluating for everything. So, caring for patients with aneurysmal subarachnoid hemorrhage obviously can be really, you know, challenging from the medical perspective, but also from the perspective of, you know, communication with families, and families asking questions about prognosis and things like that (and you mentioned this in your article about prognostication a little bit) - and can you talk a little bit about our ability to prognosticate long-term outcomes for patients who are in that acute phase (maybe even early first, you know, couple of days or a week) with a subarachnoid hemorrhage?

Dr Park: I think one of the most rewarding aspects of caring for patients with subarachnoid hemorrhage is that these patients can look, really, very sick in the beginning, and they're quite complex to manage, but you can see some very impressive recovery. And from a neurointensivist perspective, seeing that recovery in kind of a rapid timeline is rare - and we get to see that in subarach patients. We see patients who just have refractory recurrent vasospasm and delayed cerebral ischemia getting all of the tools thrown at them and you're really kind of, you know, concerned that there seems to be no end - but there is this peak of that injury, and then after that window of secondary brain injury risk kind of resolves, the patient can very much recover (so seeing patients who look the sickest be able to leave and go home). I think there is a hidden cost to subarachnoid hemorrhage where, maybe on our gross measures of outcome, patients look great, but there are this hidden cost of social psychological outcome that is unmeasured the way that we are currently measuring it. And I think our field is getting better at adopting some of the ability to measure those kind of hidden costs, and we're able to see that, even a year out, patients are really not back to where they were before (even though on the scales we currently have, they do look great, right, in terms of motor function, and things like that) - so, I think as clinicians, we have to be sensitive to that. So, when we talk to families, we have to remain hopeful that they are going to have a remarkable potential recovery but prepare families that they really should be on the lookout for any opportunity to rehabilitate in all aspects of function.

Dr Nevel: Yeah. And you mentioned in your article that as we're moving into the future - and even currently - that there is some focus on gathering more patient-reported outcomes for people who are, you know, out of the ICU back in their normal lives after subarachnoid hemorrhage (which speaks to this that you're talking about, that even if their motor function is normal, they may not be back to their normal lives). So, what is something you think that's really important that we've learned in the past ten years - I'll give it ten years, you can go back further, make that time frame shorter if you want, but about the past ten years - about subarachnoid hemorrhage's impact on patient care, and then what do you think we're going to learn in the next ten years that will impact the way we care for these patients?

Dr Park: So, you know, subarach - in terms of the literature that is forming, that has formed - like I said, the guidelines had not been updated for over a decade, and we're fortunate to have not just one, but two sets of guidelines from two professional societies that were published right next to each other this past year in 2023 - but the field is fast moving, so even after the publication of those guidelines, there was one of the first randomized controlled trials in the field to be published maybe a month or two after that (that was the early lumbar drain trial). So, the key areas that I think where the literature has really helped strengthen our practice in terms of bringing standardization is in the antifibrinolytics. And so, in that space, recently, there was a very nicely performed randomized controlled trial for early administration of antifibrinolytics. It's a practice that, even when I was training, was sort of based on old literature back when we used to treat subarachnoid patients very differently - so we were really kind of extrapolating from that literature into our practice, and we were all sort of just giving it uniformly to patients early on with the good intention to try to prevent rebleeding, (which we understood, prior to aneurysm securement, was a high source of morbidity/mortality). So, in trying to reduce that risk of rebleeding (which happens very early) as much as we could, we were giving it. But the length of treatment (you know, who should we give that medication to) was really kind of uncertain - and this recent randomized controlled trial really gave a definitive answer to this, which is that it probably makes no difference. It should be seen with a caveat, though, that the trial (like any trial) was a very specific population. So, it could probably be said that for patients who are secured very early, there's no role for antifibrinolytic therapy, but, potentially, for patients who may be in a lower-middle-income environment or lower-income environment or for whatever reason can't reach aneurysm securement within that seventy two-hour period - you could consider, you know, greater than twenty-four hours you should consider the use of antifibrinolytics - but largely has brought an end to uniform administration of antifibrinolytics. This is where that expert nuanced care comes to, right?

Dr Nevel: Mm-hmm.

Dr Park: Another area is, really, kind of something as basic as blood pressure management. I think we were taught very early on that we should be very rigorous, bring that blood pressure down - and so, I think, across all types of stroke now, we're realizing there is a little bit of nuance, right? You have to think about your patient, about prior existing renal failure, about prior existing chronic hypertension that's poorly controlled - and in subarachnoid hemorrhage, the additional impact of that early brain injury. If you have cerebral edema, you should be considering, do we really want to control our blood pressure that low? Because we might be inducing secondary brain injury from our presumed protective intervention. So, these types of things are being revisited - so, the language around that in the new guidelines is a little bit softer, and it does sort of refer more to, "let's consider the whole patient".

Dr Nevel: Yeah, rather than making a blanket statement that doesn't apply to maybe everybody.

Dr Park: Yeah. And you also asked about future.

Dr Nevel: Yeah. Where do you think things are heading in the future? What's exciting in research, and if you had a crystal ball, what do you think we're going to figure out in the next ten years that's going to impact care?

Dr Park: Well, fortunately, for patients with subarachnoid hemorrhage and for people like me who are treating patients with subarachnoid hemorrhage, there's a lot going on. So, I mentioned lumbar drainage because there was a very nice trial that was published - I think we'll see in the next few years how much of that diffusion of innovation travels across the country in the world about the usage of this. There are some who point to prior studies that may have conflicting results and so want to wait and see it be validated. Others are pretty convinced, you know, by the quality of the study that was done and are trying to incorporate it into their protocols now. I think we're going to see more usage and more study of things like intravenous milrinone, early stellate ganglion blockade, intraventricular nicardipine, and even maybe optimized goals for cerebral perfusion or blood pressure - and this is for looking at a myriad of outcomes, including the prevention and treatment of vasospasm and ischemia, improving outcomes, and preventing infarction. There's also a lot to come about early brain injury (and I kind of talked about that). It's like a seventy-two-hour period window after subarachnoid hemorrhage, and it comprises processes like microcirculatory dysfunction, blood-brain barrier breakdown, and things like oxidative cascades, et cetera. While currently, there doesn't exist any practice besides, like, the nuance and expert determination of blood pressure goals prior to aneurysm securement, I think this will be an area that hopefully will become a target for intervention, because it has an independent and influential impact on poor outcomes for subarachnoid hemorrhage patients. So, watch the space.

Dr Nevel: Yes, absolutely. Looking forward to seeing what comes. Well, thank you so much for talking to me, Dr Park, and joining me on Continuum Audio.

Dr Park: It was my pleasure.

Dr Nevel: Again, today, I've been interviewing Dr Soojin Park, whose article on emergent management of spontaneous subarachnoid hemorrhage appears in the most recent issue of Continuum in neurocritical care. Be sure to check out Continuum Audio episodes from this and other issues. And thank you to our listeners for joining today.

Dr Monteith: This is Dr Teshamae Monteith, Associate Editor of Continuum Audio. If you've enjoyed this episode, you'll love the journal, which is full of in-depth and clinically relevant information important for neurology practice - and right now, during our spring special, all subscriptions are 15% off. Go to Continpub.com/Spring2024 or use the link in the episode notes to learn more and take advantage of this great discount. This offer ends June 30, 2024. AAN members, go to the link in the episode notes and complete the evaluation to get CME. Thank you for listening to Continuum Audio.