Announcer:
0:00
Welcome to MedEvidence!, where we help you navigate the truth behind medical research with unbiased, evidence-proven facts. Hosted by cardiologist and top medical researcher, Dr Michael Koren.
Dr. Michael Koren:
0:11
Hello, I'm Dr. Michael Koren, the executive editor of MedEvidence! And if you ever wondered what AI and psychiatry and clinical research had in common, you may find this podcast to be fascinating. Now, quite frankly, I don't think anybody has thought about those three things in the same sentence, but I have Dr. Rachna Saralkar here and she and I are going to talk about that the intersection of artificial intelligence, psychiatry and clinical research. So, Rachna, welcome to MedEvidence! Thank you for joining me, and I'm really excited about this conversation. I think it's going to take off in many different directions, so we'll have some fun with this and hopefully do a little education along the way.
Dr. Rachna Saralkar:
0:54
That sounds great. Thanks, Michael, I'm glad to be here.
Dr. Michael Koren:
0:58
So just to start us off, let our MedEvidence audience know about your background, how you got to become a principal investigator and your journey. That covered all those areas that I just spoke about.
Dr. Rachna Saralkar:
1:10
Yeah, absolutely. So I'd say. My journey started off in medical school at Thomas Jefferson University, then I went on to do my psychiatric residency at Johns Hopkins After graduating from residency-
Dr. Michael Koren:
1:23
We have a little rivalry there. As a Harvard guy, sometimes you don't always get along.
Dr. Rachna Saralkar:
1:27
Oh, no, yes.
Dr. Michael Koren:
1:29
But I like you, so I'll make an exception.
Dr. Rachna Saralkar:
1:32
I appreciate you letting me be on the show, no, so after that I moved out to the Bay Area for almost 10 years and while I was out there I wanted to really focus on patient care. I wanted to make sure that I was going to be a really good doctor. You know I was. I wanted to make sure in psychiatry we see patients in all sorts of different areas. So in psychiatric emergency rooms I worked in inpatient units, I was a consult liaison psychiatrist in the hospital and I also held an outpatient practice for 10 years. So, yeah, I wanted to make sure I was good at delivering care and have real experience in that.
Dr. Rachna Saralkar:
2:19
And then about three years ago I was really frustrated by the amount of variability I saw in the quality of care patients got before they came to see me, depending on the different types of clinicians that they had seen, the types of treatments they had received. And the second thing that really bothered me were how poor the outcomes were, how poorly patients were really doing, especially the ones who felt like they had tried everything and spent years trying meds and treatments before they finally found one thing that worked. And in medicine we call this kind of "stratification and we wish in psychiatry that we had better stratification to pick the right treatment for the right person first, and I was really disheartened by the lack of that and I said I want to go back to school and figure out how I can be part of finding some solutions to these problems.
Dr. Michael Koren:
3:07
Very, very cool Very cool and so
Dr. Rachna Saralkar:
3:09
Yeah, go ahead.
Dr. Michael Koren:
3:11
Yeah, so most recently you're a principal investigator now with a national company that runs clinical trials, so talk a little bit about that connection.
Dr. Rachna Saralkar:
3:18
Yeah, so I I ended up being here at Flourish Research as an investigator, but the way I landed here, the way I got interested in research, was after I was at a startup for the last couple of years working on using AI to measure depression levels and anxiety levels using audiovisual features. So the way somebody's face moves, the way their voice changes when they're talking, and during the process of being in that world of working with the FDA, attending conferences, seeing other companies roll out AI therapists and lots and lots of different tools, I was really frustrated by the quality of evidence that backed a lot of these tools and I said I want to be an expert in clinical research and I want to know how to properly validate treatments and tools in an ethical manner, because I want my patients getting well-validated care at the end of the day, and so that's why I entered clinical research.
Dr. Michael Koren:
4:27
Interesting.
Dr. Michael Koren:
4:28
So just educate me. This is not an area that I really track. How far is AI in treating a patient with depression? Is an AI bot as good or better than a therapist at this juncture?
Dr. Rachna Saralkar:
4:42
At this juncture, it is definitely not even close to being better than a therapist, I'd say.
Dr. Rachna Saralkar:
4:47
You know, I think there's a lot of really interesting studies and a lot of really great tools that are coming out, and some of them show really promising data that you know. AI tools and AI therapists probably can truly emote empathy and they probably really can create patient alliance between the AI bot and the patient, and so there are studies showing that this can really happen and I'd say, when patients know who they're talking to and they're okay with the fact that they're talking to an AI agent, an AI bot, then it can work really well. There's a lot of dangers associated with it. Still, I'd say there are a lot of problems that have not been ironed out a lot of safety issues, privacy issues and I'm happy to get into the other things that I think are of concern there but we're really getting closer and closer to having AI at least be part of the therapeutic process, right? So I think there's a difference between an AI bot being a therapist themselves versus AI being part of the process and trying to improve the treatments.
Dr. Michael Koren:
5:57
Interesting. Yeah, I've heard that stated before about AI bots having empathy, and I have to say that I find that a statement that I'm going to push back on, for the simple reason that empathy is intrinsically human is. A computer can't have empathy. They can fake empathy, they can make the recipient think there's empathy, but empathy really, sort of by definition of the word, requires two human beings.
Dr. Rachna Saralkar:
6:33
Yeah.
Dr. Michael Koren:
6:34
So yeah.
Dr. Rachna Saralkar:
6:36
I totally agree with you. I think this comes into the problem of, I think of language, in my opinion. So we have this word, empathy, and it's the word that we've had for a long time. So that's how we are trying to describe this relationship between doctor and patient, and we're trying to, and there are scientists who are studying what the factors that go into empathy and trying to measure well, how well do AI bots do that? When? To your point, I completely agree with you.
Dr. Rachna Saralkar:
7:03
If I met an alien and a human being on a new planet for the first time, I don't care how well the alien talks to me, I know that the human being understands me and feels the same feelings that I feel and has the ability at least to feel those things, and that immediately creates a connection that an alien or a robot or an AI agent could never do. I cannot form that connection with them, but I do think I can feel something with that alien. I do think I could feel something with that bot. I just don't think we have a term for it yet.
Dr. Michael Koren:
7:35
Right, right, yeah, it's interesting. Obviously, as humans, we need to know that other people care about us. We need to know that our suffering is not unique to us. I think those are important things, but how to get that message across is really what sounds like. Your research is involved, and I had an interesting conversation recently. I think it was meant as an insult to me, but I was talking to a patient and she mentioned to me that she got a lot of solace from her dog, that her dog paid attention to her, and she never questioned whether or not somebody that she was talking to was paying attention to her when it was with her dog, like the feelings that she had with human beings. And then she was talking to me about it. I didn't know if she was accusing me of this or not, but especially in the modern era where we're looking back and forth from our EMR and the people are wondering are you really paying attention to me?
Dr. Rachna Saralkar:
8:32
Yeah, absolutely.
Dr. Michael Koren:
8:33
And she made a specific point that when she was interfacing with the healthcare system she wasn't getting that, but when she interfaced with her puppy she got that. So you know, interesting. And then the other thing I'll throw out just to be a little provocative about quote computer programming and human intelligence, is how long it took computers to beat humans in chess. I don't know if you've studied that or not, but it sounds like you probably have. But just for the listeners, IBM started looking at this problem back in the very early 1960s and they were doing this on mainframe computers that were still very, very powerful and they were having the hardest time, despite having dozens of computer scientists working on this problem of beating a grandmaster in chess, because the human brain is really good at pattern recognition and basically they would see a pattern and the humans would know how to manipulate that pattern.
Dr. Michael Koren:
9:29
The computers were always catching up, and so for a long, long time it was actually over 30 years computers could not beat the best humans in chess. And then all of a sudden they changed the programming to make the computers more human, which was to start bluffing. So the computer started to learn how to just make a crazy move that made no sense that was something that a grandmaster hadn't seen before, and at that point the grandmasters would panic a little bit. They wouldn't know exactly how to deal with it. And then there was an opening for the computers to win, and nowadays computers can absolutely destroy humans in chess. So there's no grandmaster that can consistently beat a computer. It's really flipped.
Dr. Michael Koren:
10:10
But it took so many years to do that, and in psychiatry that gets even more complex than chess.
Dr. Rachna Saralkar:
10:16
Oh, it's so much more complex, but I mean, I think what's for me, what was really kind of uplifting and exciting about the entire history of AI and chess was that at the end of those 30 years now the grandmaster you know. Before it used to take, you know, many, many years for somebody who was like an apprentice in chess to become a grandmaster. That time, I think, has been cut in half because now they're training with AI, they're practicing against AI, and how good they get is so much faster and that's what I'm hoping is happening in healthcare right.
Dr. Michael Koren:
10:54
I love that point. That's a brilliant point. Thank you for bringing that up. That's exactly right. So sometimes you have to think about things from the other way around and I absolutely love that. There's so many places where the tail wags the dog and we always have to think let's get the dog to wag the tail.
Dr. Rachna Saralkar:
11:11
Yeah, yeah, no, and I definitely think that's where the most important applications will be in healthcare. You know it doesn't sound as sexy or as fun, but education using AI and helping doctors become better doctors, therapists become better therapists using AI is out there and that's really exciting.
Dr. Michael Koren:
11:33
Yeah, that's very cool, yeah. So switching gears a little bit, tell me about some of the areas that you're working on now in terms of clinical research. What are you doing, what areas are you looking to do in the future, and a little bit about the patient value proposition. I think one of the things that I really focus on as a cardiologist that does a lot of clinical trials is how do I bring value to my patients? So maybe you can comment on that in your realm.
Dr. Rachna Saralkar:
11:59
Sure. So our site in Philadelphia. We have a lot of Alzheimer's trials right now and we're growing out our depression trials at the moment and I think that both of these sets of trials are so I think are so important. You know, we have a lot of connections within our community to groups of foundations and group homes and places where we know, you know, a geriatric demographic lives and we do our best to go out to those places and form connections in those communities so we can educate everyone.
Dr. Rachna Saralkar:
12:38
You know, even if we only get one or two people who end up being interested in a clinical trial, I think it's so important for people to know that we're at the precipice of being able to treat things that maybe you only know exist in your family history and you aren't even having symptoms of yet, like Alzheimer's disease, right. In the depression realm,
Dr. Rachna Saralkar:
12:57
I think there are so many, you know, cutting edge treatments that are, you know, one trying to cut down on some of the horrible side effects of the existing treatments that we have and two, trying to treat treatment resistant depression in a way we've never been able to do before, and I think those are really exciting areas that both of those things I think aren't really out in the market yet. And when patients come here, I love my job because I get to spend as much time as they need going through and educating them about not just what trials we have but what the history of the disease processes that run in their family are and what potential treatments we could offer or even they could get from their outside doctor.
Dr. Michael Koren:
13:45
Yeah, it's fascinating, so let's dig into that a little bit more. I think those are both super interesting areas. We'll start with depression, and again, I'm a cardiologist, but our site here in Florida has been involved in depression studies in the past. In fact, one of the studies that I was personally involved with was called the SAD Heart Study, which looked at Sertraline back in the day as a way of treating post-myocardial infarction depression, post-heart attack of depression.
Dr. Rachna Saralkar:
14:13
Wow fascinating.
Dr. Michael Koren:
14:14
Yeah, it turned out to be a very successful and safe treatment, most importantly safe post-MI, and probably is life-saving in many cases. But the problem with those studies is that there is a very strong placebo effect. So when you do a placebo-controlled study in depression, just the interaction with the team and the staff and the physicians seems to help people.
Dr. Michael Koren:
14:38
Again, this empathy concept that we're talking about when we started so talk to us a little bit about that and ways of dealing with that, knowing it's not necessarily a bad thing At the end of the day if the study itself helps people that's great.
Dr. Rachna Saralkar:
14:54
Yeah, yeah.
Dr. Rachna Saralkar:
14:55
No, I think it's important for patients to understand that when you walk into someplace that's doing clinical trials and you're there for depression, it's different from walking in to see your psychiatrist or your therapist, because when there is a trial in depression to your point, the placebo effect means that people could get improvement from not from just the interaction and help from talking to people in the community, talking to people at the site. And so what we do here is we do our very best to treat everybody exactly the same, which is not what we do really in personalized one-on-one therapy care. Right, we would dig into each of your problems each day and dive into what's bothering you and what's not. We won't do that at a clinical trial site, because we want to treat everybody the same with kindness, with empathy, and we always will, and we can provide education, but we try not to. We don't go as deep into kind of emotional issues.
Dr. Rachna Saralkar:
15:57
For that reason.
Dr. Michael Koren:
15:58
Right, right, but at the end of the day you know part of us it plays a scientific role and we want to see if the investigational product makes a difference.
Dr. Rachna Saralkar:
16:09
Yeah, yeah.
Dr. Michael Koren:
16:10
But part of us, as good clinical investigators, are clinicians. We're physicians that want to see our patients get better, regardless of their assignment in a research protocol, and so from that standpoint we want to encourage people A to get into clinical research, because even people that get placebo seem to benefit from the interaction, and that's okay. It makes it harder scientifically in some places, but again, as a clinician to clinician, we're okay with that concept.
Dr. Rachna Saralkar:
16:38
It's what my field relies on completely. You know, I am more than okay with the placebo effect and I think as long as we're, you know, and most patients are too, you know.
Dr. Rachna Saralkar:
16:48
so I agree
Dr. Michael Koren:
16:50
Yeah, and that's also true not only in psych studies but other things that you would. You would think would be more objective, but they actually are driven by stress, anxiety and psychological issues. The number one that I've been involved with is blood pressure. Is that when you do a placebo controlled blood pressure study, people's blood pressure goes down on placebo just because they have the confidence of developing this rapport with the team and probably complying better with their baseline diet and exercise and everything else, and things all move in the right direction. And I can tell some stories about really promising technologies that were very difficult to prove because of the placebo effect of people being in a blood pressure study.
Dr. Rachna Saralkar:
17:32
That's amazing. I didn't know that.
Dr. Michael Koren:
17:34
Yeah, so interesting stuff how that translates to different areas in clinical research. So let's shift gears a little bit to Alzheimer's or memory issues, and you brought up a really, really interesting point is we don't know exactly when to intervene for this disease, and I think it's fair to say that most of the interventions that have been put out there to date have looked at the end stages of disease rather than treating earlier. So maybe comment a little bit on that and help us understand that balance and trade off a little bit more.
Dr. Rachna Saralkar:
18:07
Yeah, sure. So a lot of the earlier studies to your point we're looking at. You know, if somebody already has Alzheimer's disease, what can we do? And so there are, you know, a few FDA approved medicines that help pull one of those proteins that can lead to Alzheimer's called amyloid, out of the brain. What they found was that, you know, when you do that so late stage to somebody who's already having significant symptoms one, it only seems to reduce the symptoms about maybe 20 to 30% at best. And second, there's more side effects from the medication versus when we start treating earlier, when there's less of that amyloid and tau protein that can lead to the disease, but there is still some. There seem to be fewer side effects and we can get people significantly better outcomes because we stop cell death, we stop your neurons from dying early on. So I think this research is so important and it's going to save millions of lives.
Dr. Michael Koren:
19:06
Yeah, yeah, obviously Alzheimer's disease is a huge problem, especially as the population ages and is less likely to die of a heart attack or stroke. You're more likely to suffer from these other problems, so we're seeing that epidemiologically. But getting back to your interest in AI, is there an AI bot that will make Alzheimer's go away? Or if you do enough puzzles, you don't get the Alzheimer's in the first place? Is this being studied critically and properly?
Dr. Rachna Saralkar:
19:32
I don't know, I have no idea. I don't know if there's any. I'm sure somebody somewhere is doing something with AI. I mean the most I know there are a lot of companies that are building things out for, yes, for cognitive learning and cognitive testing using AI, and I do think those tests are great but probably need to be well validated, which was kind of, again, one of the big reasons why I came into clinical research, because I think all those things need to be better validated.
Dr. Michael Koren:
20:01
Yeah, maybe you and I should write an investigator-initiated study looking at an intervention of doing puzzles using an AI bot or maybe chess or checkers we were talking about that and then randomizing people at risk and seeing how "smart they are quote, unquote two years down the road and whether or not they have any, how well they do on validated testing of memory. That might be an interesting little project. That might be very interesting. Well, we'll put that on our list of to-dos.
Dr. Rachna Saralkar:
20:35
Sounds good.
Dr. Michael Koren:
20:37
And yeah, Rachna, this has been a fabulous conversation. I don't know if there's anything else you want to present to the audience in terms of some of the challenges or things that you see for the future with regard to clinical research in your new role.
Dr. Rachna Saralkar:
20:51
You know, I think the only thing I would add and it's probably a larger conversation, something that I am starting to see when I'm going through these protocols with patients is that when there are requests for data, requests for scans and images that a company is asking for, I think we're working on making doing a better and better job of making sure that patients understand the risks that are involved with that, not just today, but could be down the line with the data, because I think that's a fear that a lot of people who kind of sign away data rights have that you know, what are people going to be able to do with my data? We're moving into a world where we're videotaping audiotaping in clinical trials sometimes and we want to, and it's really important to me that patients understand and that we're being careful about the protocols we choose, that patients aren't ever being limited by the testing that we're doing.
Dr. Michael Koren:
21:58
Yeah, I love that point. It's a fabulous point and certainly very, very important. Well, Rachna, this has been a delightful conversation. Thank you for being part of MedEvidence! and I look forward to working with you in the future.
Dr. Rachna Saralkar:
22:10
Same Thanks so much.
Announcer:
22:11
Thanks for joining the MedEvidence! podcast. To learn more, head over to MedEvidence. com or subscribe to our podcast on your favorite podcast platform.
