WEBVTT NOTE This file was generated by Descript 00:00:00.029 --> 00:00:04.460 Judith: Welcome to Berry's In the Interim podcast, where we explore the 00:00:04.460 --> 00:00:08.340 cutting edge of innovative clinical trial design for the pharmaceutical and 00:00:08.340 --> 00:00:10.760 medical industries, and so much more. 00:00:11.010 --> 00:00:11.739 Let's dive in. 00:00:12.835 --> 00:00:13.465 Scott Berry: All right. 00:00:13.465 --> 00:00:14.635 Welcome everybody. 00:00:14.635 --> 00:00:20.215 Back to, in the interim, I'm your host, Scott Berry, and we are going 00:00:20.215 --> 00:00:26.245 to dive into a number of things of Bayesian adaptive trials, the life in 00:00:26.245 --> 00:00:30.715 the academic world a little bit today I have, uh, a guest with me today. 00:00:30.715 --> 00:00:32.100 I have Dr. 00:00:32.430 --> 00:00:33.460 Byron Eski. 00:00:34.350 --> 00:00:38.940 a professor of biostatistics and data science at the University of Kansas 00:00:38.940 --> 00:00:44.100 Medical Center and the University of Kansas Cancer Center, uh, in Kansas 00:00:44.100 --> 00:00:49.770 City, and he's interested in Bayesian modeling, Bayesian adaptive trials, 00:00:50.040 --> 00:00:56.130 and he's a fellow of the American Statistical Association and he's 00:00:56.130 --> 00:00:57.570 also a very good friend of mine. 00:00:58.230 --> 00:01:00.240 Byron, welcome to in the interim. 00:01:01.795 --> 00:01:04.135 Byron Gajewski: Scott, uh, thanks for having me on this. 00:01:04.215 --> 00:01:06.395 I'm a, I'm a big fan of the pod. 00:01:07.095 --> 00:01:13.055 Uh, this in the interim and, um, Rewatchables is another 00:01:13.055 --> 00:01:16.855 podcast that I listen to quite often, both of them. 00:01:16.955 --> 00:01:19.535 I don't know if you know about Rewatchables. 00:01:19.695 --> 00:01:23.115 It's, it's produced and done by Bill Simmons as a sports guy 00:01:23.530 --> 00:01:23.890 Scott Berry: Ah, 00:01:24.295 --> 00:01:25.475 Byron Gajewski: and he does, 00:01:25.895 --> 00:01:30.055 yeah, oh you do, and he does, uh, Rewatchable movies. 00:01:30.345 --> 00:01:31.365 That's what he does. 00:01:31.445 --> 00:01:32.225 He talks about it. 00:01:32.285 --> 00:01:35.445 So I listen to both of them, uh, as I, as I 00:01:35.445 --> 00:01:37.265 walk or, or, or whatnot. 00:01:37.845 --> 00:01:38.585 Um, exercise. 00:01:40.030 --> 00:01:43.840 Scott Berry: so is a re watchable movie a good movie? 00:01:44.170 --> 00:01:48.280 Uh, maybe really good movies you don't want to rewatch necessarily. 00:01:48.425 --> 00:01:51.585 Byron Gajewski: It's interesting because, uh, but the concept is, 00:01:52.065 --> 00:01:55.165 is you're sitting there and, and we're cable, I'm a cable person. 00:01:55.245 --> 00:01:58.165 Or in the old days, regular TV and antennas and stuff. 00:01:58.485 --> 00:02:00.005 And a movie just happens to be on. 00:02:00.865 --> 00:02:02.905 And you're taking a nap and you look up and you go, oh 00:02:02.905 --> 00:02:04.285 gosh, that's a great movie. 00:02:04.285 --> 00:02:04.465 Okay. 00:02:04.650 --> 00:02:07.110 I really want to watch where I'm at right now. 00:02:07.170 --> 00:02:09.010 You don't have any streaming or anything like that. 00:02:09.050 --> 00:02:10.370 So it's a, it's a rewatchable. 00:02:10.450 --> 00:02:14.090 It's not necessarily the greatest film, but they 00:02:14.090 --> 00:02:15.250 usually are good films. 00:02:15.250 --> 00:02:17.010 And they're, they're just very rewatchable. 00:02:17.570 --> 00:02:17.770 yeah, 00:02:18.510 --> 00:02:19.730 sure. 00:02:19.855 --> 00:02:22.255 Scott Berry: hard to turn them off and it's hard to get up from. 00:02:22.255 --> 00:02:22.525 Yeah. 00:02:22.555 --> 00:02:22.795 Yeah. 00:02:22.795 --> 00:02:25.675 And, and you gotta kind of watch it, despite the fact you've seen 00:02:25.675 --> 00:02:27.565 the movie six or seven times. 00:02:27.625 --> 00:02:27.835 Yeah. 00:02:28.150 --> 00:02:28.590 Byron Gajewski: Absolutely. 00:02:29.110 --> 00:02:31.330 Uh, you can't handle the truth. 00:02:31.430 --> 00:02:34.290 I say that and you know, immediately, or at least some 00:02:34.290 --> 00:02:36.750 people will know immediately what that rewatchable movie is. 00:02:36.770 --> 00:02:37.330 If you couldn't. 00:02:38.050 --> 00:02:39.670 Scott Berry: Yeah, this just happened a couple 00:02:39.710 --> 00:02:40.150 Byron Gajewski: Oh 00:02:40.360 --> 00:02:45.280 Scott Berry: Tammy has a gift and that is, she doesn't remember movies. 00:02:45.730 --> 00:02:46.830 Byron Gajewski: yeah, that is, yeah. 00:02:47.050 --> 00:02:51.970 Scott Berry: Tammy, and so she can watch a movie and watch the same movie 00:02:51.970 --> 00:02:55.570 six months later and be in complete suspense as to what's gonna happen. 00:02:55.570 --> 00:02:56.230 'cause she doesn't 00:02:56.410 --> 00:02:56.970 Byron Gajewski: Yeah, 00:02:57.130 --> 00:02:57.460 Scott Berry: happen. 00:02:57.880 --> 00:02:58.780 Yes, yes. 00:02:58.810 --> 00:02:59.590 It it, it's a 00:02:59.770 --> 00:03:02.710 Byron Gajewski: that is a great gift, actually. 00:03:03.325 --> 00:03:03.745 Scott Berry: Yeah. 00:03:04.105 --> 00:03:04.525 Yep. 00:03:04.675 --> 00:03:05.065 Yep. 00:03:05.150 --> 00:03:05.710 Byron Gajewski: Sure. 00:03:05.845 --> 00:03:06.295 Scott Berry: Okay. 00:03:06.295 --> 00:03:13.315 So Byron, uh, let's, let's, um, get into some statistics, some adaptive trials. 00:03:13.315 --> 00:03:18.235 Bayesian, of all, and I, I know you watched the podcast, uh, with 00:03:18.235 --> 00:03:22.825 Sean Cassidy, and he, he describes, you know, tell me your story. 00:03:23.995 --> 00:03:27.835 let, let's, let's let everybody know what is the Byron Eski story? 00:03:28.710 --> 00:03:32.150 Byron Gajewski: Oh, yeah, that's a, that's a good question. 00:03:32.810 --> 00:03:36.250 Um, you know, uh, maybe I could go into some of my background 00:03:36.250 --> 00:03:40.730 academically, maybe a little bit about how that went, how 00:03:40.730 --> 00:03:42.970 I was trained in statistics. 00:03:42.970 --> 00:03:47.870 I studied math and civil engineering at Marquette University, bachelor's 00:03:47.870 --> 00:03:49.370 degree, master's degree in math. 00:03:49.435 --> 00:03:53.455 Um, my dad was a civil engineer. 00:03:54.220 --> 00:03:54.300 Scott Berry: Hmm. 00:03:54.575 --> 00:03:57.355 Byron Gajewski: I loved math, so I studied civil engineering. 00:03:57.455 --> 00:04:01.275 Funny story on that, I was happy to be in the College of 00:04:01.275 --> 00:04:04.335 Engineering at Marquette, uh, because I didn't have to go to 00:04:04.335 --> 00:04:07.755 arts and sciences and do, I, I thought back then, too much writing. 00:04:08.675 --> 00:04:08.975 You know, 00:04:09.375 --> 00:04:12.555 it was hard for me to be in arts and sciences at the time. 00:04:12.655 --> 00:04:16.595 It was, so, so anyway, I studied engineering and math there. 00:04:16.595 --> 00:04:16.735 It was hard for me to be in arts and sciences at the time. 00:04:16.735 --> 00:04:19.975 And then I studied math in master's, and I realized that 00:04:19.975 --> 00:04:24.575 math was too theoretical for me, for my taste to do forever, 00:04:25.515 --> 00:04:26.935 and engineering was too applied. 00:04:27.375 --> 00:04:29.135 Statistics was kind of a sweet spot. 00:04:29.915 --> 00:04:34.215 I had an advisor in math, who, who, I had the book, back then you 00:04:34.215 --> 00:04:38.975 look at graduate programs in, in statistics, and you had a big book 00:04:38.975 --> 00:04:41.695 you pulled out of the library, and you said, hey, what do you think? 00:04:41.695 --> 00:04:41.985 And I was like, okay. 00:04:42.705 --> 00:04:44.625 And he starts looking around. 00:04:44.905 --> 00:04:48.365 He's a bio -math guy, statistician in a math department. 00:04:50.065 --> 00:04:51.825 And he said, Texas A &M. 00:04:52.545 --> 00:04:55.045 He said, they're a good program and they got money. 00:04:57.545 --> 00:05:00.605 And it turns out they had a pot of money to bring prospective 00:05:00.605 --> 00:05:01.645 graduate students down there. 00:05:01.685 --> 00:05:03.685 I went and visited and fell in love with Texas A &M. 00:05:04.285 --> 00:05:05.045 Loved it down there. 00:05:05.365 --> 00:05:09.965 And that's where I studied for my PhD in statistics at A &M. 00:05:10.120 --> 00:05:14.500 And it was funny because, um, at A &M at the time, 00:05:14.820 --> 00:05:18.480 it was a very frequentist department, very frequentist. 00:05:19.020 --> 00:05:21.680 Uh, but, um, and I studied. 00:05:21.740 --> 00:05:23.500 I remember it being a very rigorous program. 00:05:23.800 --> 00:05:27.480 I studied everything from statistical computing to 00:05:27.480 --> 00:05:29.240 measure theoretic probability. 00:05:29.805 --> 00:05:30.225 Scott Berry: Mm-hmm. 00:05:30.260 --> 00:05:32.080 Byron Gajewski: Uh, the exams were rigorous. 00:05:32.660 --> 00:05:34.320 Um, I did take a base course there. 00:05:34.880 --> 00:05:36.540 I did take a base course. 00:05:36.640 --> 00:05:37.820 I really liked it. 00:05:37.820 --> 00:05:42.805 Um, You talk about in a previous pod, Jimmy Savage, 00:05:43.625 --> 00:05:47.725 uh, he co -authored a paper that I read, actually my professor assigned 00:05:47.725 --> 00:05:50.605 it to me from Psychological Review, 00:05:51.525 --> 00:05:56.485 um, a foundational paper in Bayesian, and I, I, I, it was really cool. 00:05:56.965 --> 00:05:58.605 You know what though, Scott, I, I just, 00:05:58.645 --> 00:06:01.555 Scott Berry: probably Edwards, Lindeman and Savage. 00:06:01.845 --> 00:06:06.925 Byron Gajewski: yes, absolutely, um, and I actually had a student 00:06:06.925 --> 00:06:09.855 read that here, um, Uh, recently, 00:06:10.495 --> 00:06:12.375 um, a foundational paper in the field. 00:06:13.135 --> 00:06:19.655 And, um, I, I took the course, loved it, but I, I didn't 00:06:19.655 --> 00:06:20.615 really use it right away. 00:06:20.855 --> 00:06:22.515 I didn't use Bayesian stuff right away. 00:06:22.635 --> 00:06:24.215 But later in my career, 00:06:24.280 --> 00:06:25.090 Scott Berry: so just going 00:06:25.235 --> 00:06:26.155 Byron Gajewski: it 00:06:26.260 --> 00:06:28.270 Scott Berry: professor, uh, in that course? 00:06:28.535 --> 00:06:29.635 Byron Gajewski: was, uh, you! 00:06:30.240 --> 00:06:30.370 Scott Berry: Yeah. 00:06:31.690 --> 00:06:31.810 Yes. 00:06:31.870 --> 00:06:32.080 So 00:06:32.095 --> 00:06:33.235 Byron Gajewski: Uh, 00:06:33.670 --> 00:06:35.950 Scott Berry: good to let everybody know that, uh, yes. 00:06:36.130 --> 00:06:39.880 Uh, when I was teaching at Texas a and m, Byron was a student there, a graduate 00:06:39.880 --> 00:06:44.050 student there, and I taught a Bays course, which that course was awesome. 00:06:44.140 --> 00:06:47.980 The, the students we had in that course and a number of them out 00:06:48.010 --> 00:06:49.990 doing things in the world of adaptive 00:06:50.455 --> 00:06:50.955 Byron Gajewski: yeah. 00:06:51.940 --> 00:06:52.450 Scott Berry: and, yep. 00:06:52.930 --> 00:06:53.800 Fantastic. 00:06:53.950 --> 00:06:54.160 Okay. 00:06:55.115 --> 00:06:59.015 So, so you take the course, it's interesting, but then 00:06:59.015 --> 00:07:00.280 where, where does it take you? 00:07:02.325 --> 00:07:05.065 Byron Gajewski: just to talk a little bit about that, I mean, 00:07:05.145 --> 00:07:07.785 my colleagues, as you mentioned, that were in that, that course. 00:07:08.145 --> 00:07:11.045 By the way, these were colleagues that, that we played pickup 00:07:11.045 --> 00:07:14.065 basketball with on Wednesday, which was a wonderful thing. 00:07:14.125 --> 00:07:15.845 And you remember playing those pickup games? 00:07:16.345 --> 00:07:20.865 It was a great way to get, blow off steam about the rigors 00:07:20.865 --> 00:07:23.925 of statistics program, but also learn about the ropes from the 00:07:23.925 --> 00:07:27.710 older, uh, older, uh, students in the group and learn about 00:07:27.710 --> 00:07:32.650 how to navigate the system, both technically and socially, and whatnot. 00:07:33.410 --> 00:07:36.210 But in that course, a lot of my colleagues, there was 00:07:36.210 --> 00:07:40.050 a number of colleagues who used Bayesian methods later. 00:07:40.730 --> 00:07:45.390 And, you know, everything from airline industry, to sports, 00:07:46.390 --> 00:07:51.290 to bowhead whale gestation age estimation, cancer. 00:07:52.745 --> 00:07:54.705 Um, and I was like, Oh, that's cool. 00:07:54.765 --> 00:07:57.245 I would like to do that, but I never really did it until 00:07:57.945 --> 00:07:59.725 I got my first job at St. 00:07:59.785 --> 00:08:00.705 Cloud State University. 00:08:01.245 --> 00:08:04.245 And in the summers I did some, some Bayesian modeling. 00:08:04.385 --> 00:08:08.365 I did some modeling where I had this, I remember vividly there 00:08:08.365 --> 00:08:12.025 was a problem where we had, it was an audiology problem. 00:08:13.120 --> 00:08:13.360 Scott Berry: Hmm. 00:08:14.345 --> 00:08:18.665 Byron Gajewski: I laugh because the experiment was on guinea pigs. 00:08:18.665 --> 00:08:19.445 Okay. 00:08:19.700 --> 00:08:24.320 Um, back to my point about my struggles with writing. 00:08:24.910 --> 00:08:25.210 Scott Berry: Yeah. 00:08:26.260 --> 00:08:30.320 Byron Gajewski: Um, I, I remember calling them pigs 00:08:30.320 --> 00:08:31.520 when I talked about it. 00:08:31.880 --> 00:08:34.160 And I couldn't pronounce guinea for some reason. 00:08:34.320 --> 00:08:36.560 I didn't, guinea didn't make sense to me. 00:08:37.320 --> 00:08:38.720 Um, and my wife makes fun of me. 00:08:38.780 --> 00:08:40.600 My wife Mary, she, she says, pigs? 00:08:40.720 --> 00:08:42.040 I thought you said they were pigs. 00:08:42.560 --> 00:08:46.920 Anyway, they did experiments on guinea pigs and it had 00:08:46.920 --> 00:08:48.640 to do with hearing, audiology. 00:08:48.640 --> 00:08:48.800 Anyway, they did experiments on guinea pigs and it had 00:08:48.800 --> 00:08:48.850 to do with hearing, audiology. 00:08:48.850 --> 00:08:52.200 In audiology, you have interesting interval censoring, 00:08:52.380 --> 00:08:53.520 the way they do the testing. 00:08:54.240 --> 00:08:55.800 They have interesting right censoring. 00:08:56.600 --> 00:08:59.500 And, uh, the data tends to be non -normal. 00:08:59.760 --> 00:09:01.840 It's a shifted, it turns out to be, we modeled it 00:09:01.840 --> 00:09:03.180 with a shifted log normal. 00:09:03.320 --> 00:09:05.220 It has a parameter that you estimate in it. 00:09:06.120 --> 00:09:09.520 And, boy, I struggled doing that from a frequentist standpoint. 00:09:09.700 --> 00:09:13.520 But then I did Bayesian, and it was just, it was clean. 00:09:14.140 --> 00:09:16.460 And so then from there, I just, I got hooked. 00:09:16.460 --> 00:09:17.940 I got hooked on Bayesian. 00:09:18.095 --> 00:09:19.395 I got hooked on that. 00:09:19.495 --> 00:09:20.815 I got hooked on the interpretation. 00:09:21.685 --> 00:09:21.955 Scott Berry: Hmm. 00:09:23.485 --> 00:09:23.815 Yep. 00:09:24.145 --> 00:09:24.565 Okay. 00:09:24.565 --> 00:09:25.945 So you're, you're St. 00:09:25.945 --> 00:09:26.755 Cloud State. 00:09:26.785 --> 00:09:27.955 Uh, by the way, St. 00:09:27.955 --> 00:09:30.625 Cloud is a place I near and dear to my heart. 00:09:30.625 --> 00:09:34.915 I grew up in Minnesota and I, I have a cabin that I spent time in the summer, 00:09:35.275 --> 00:09:37.435 uh, about 45 minutes from there. 00:09:37.435 --> 00:09:39.775 So, uh, fantastic place. 00:09:39.775 --> 00:09:42.175 More fantastic in the summer than the winter. 00:09:42.565 --> 00:09:43.345 Um, 00:09:43.875 --> 00:09:44.315 Byron Gajewski: Yeah, 00:09:44.365 --> 00:09:45.715 Scott Berry: where, where do you go from there? 00:09:47.195 --> 00:09:49.895 Byron Gajewski: so a brief comment on that. 00:09:49.935 --> 00:09:52.915 I was there because I liked St. 00:09:52.995 --> 00:09:53.635 Cloud State. 00:09:53.715 --> 00:09:56.075 They had a nice undergraduate statistics department. 00:09:56.195 --> 00:09:57.595 They had a program there. 00:09:57.895 --> 00:09:58.395 It was neat. 00:09:58.475 --> 00:10:01.895 I had students that went to JSM that were statistics 00:10:01.895 --> 00:10:02.815 majors in undergrad. 00:10:04.155 --> 00:10:07.595 Um, but I was there partly we had a match thing. 00:10:07.675 --> 00:10:08.375 My wife. 00:10:08.640 --> 00:10:13.340 I got a master's in social work at, at, and I'll say this to 00:10:13.340 --> 00:10:17.440 the listeners, I'll call it the U, but Americans in the Southeast 00:10:17.440 --> 00:10:21.460 might think that it's University of Miami, but it's not in Minnesota, 00:10:21.780 --> 00:10:22.980 it's University of Minnesota. 00:10:23.165 --> 00:10:23.585 Scott Berry: mm-hmm. 00:10:24.140 --> 00:10:25.920 Byron Gajewski: And you have to say it like that, Minnesota. 00:10:26.320 --> 00:10:27.700 Scott Berry: Yeah, that was pretty good, Byron. 00:10:27.700 --> 00:10:27.970 Yeah. 00:10:28.055 --> 00:10:28.345 Yeah. 00:10:29.760 --> 00:10:32.920 Byron Gajewski: And so anyway, after two years of that, we had an 00:10:32.920 --> 00:10:37.260 open market and I found I found a job at University of Kansas 00:10:37.260 --> 00:10:39.320 Medical Center here in Kansas City. 00:10:39.800 --> 00:10:42.240 Oh, by the way, there's a difference between Kansas City, 00:10:42.400 --> 00:10:44.900 Kansas and Kansas City, Missouri. 00:10:45.840 --> 00:10:47.520 Two cities, two different states. 00:10:48.560 --> 00:10:52.460 And, uh, the chiefs in Missouri now are going to be moving from Kansas 00:10:52.460 --> 00:10:54.260 City to Kansas City, by the way. 00:10:54.430 --> 00:10:54.670 Scott Berry: Yeah. 00:10:55.840 --> 00:10:56.500 Oh, so you are 00:10:56.520 --> 00:10:57.040 Byron Gajewski: Yeah, 00:10:57.250 --> 00:10:58.030 Scott Berry: Kansas, but. 00:10:58.585 --> 00:10:59.785 You live in Missouri? 00:11:00.380 --> 00:11:03.140 Byron Gajewski: I live, yeah, I live in Kansas City, Missouri and 00:11:03.140 --> 00:11:04.560 I work in Kansas City, Kansas. 00:11:05.605 --> 00:11:05.755 Scott Berry: Yeah. 00:11:05.755 --> 00:11:05.765 Yep. 00:11:06.980 --> 00:11:09.700 Byron Gajewski: Um, so, uh, we do that. 00:11:09.780 --> 00:11:11.300 And so that's how I ended up at University of 00:11:11.300 --> 00:11:13.160 Kansas, um, Medical Center. 00:11:13.300 --> 00:11:14.340 I just, an open job. 00:11:14.440 --> 00:11:15.620 My brother lived out here. 00:11:15.680 --> 00:11:16.460 I came and visited. 00:11:16.640 --> 00:11:17.300 I loved it. 00:11:17.340 --> 00:11:18.440 I loved the university. 00:11:18.600 --> 00:11:20.520 I loved what was happening here, and I still love it. 00:11:20.840 --> 00:11:21.140 I still, 00:11:21.500 --> 00:11:24.360 there's great things going on here at KUMC. 00:11:25.225 --> 00:11:27.475 Scott Berry: Now, were you doing biostatistics at St. 00:11:27.475 --> 00:11:27.595 Cloud? 00:11:30.180 --> 00:11:31.900 Byron Gajewski: I was, so I, uh, 00:11:33.125 --> 00:11:33.445 Scott Berry: I mean. 00:11:33.650 --> 00:11:36.710 Byron Gajewski: I published off of my dissertation when I was there 00:11:36.710 --> 00:11:39.930 at first, and then I got these consult, the reason I got into the 00:11:39.930 --> 00:11:44.250 guinea pig experiments was because my buddy from high school was a 00:11:44.250 --> 00:11:47.470 resident and a professor, a clinical professor at University of Florida. 00:11:47.730 --> 00:11:51.130 He, he knows me, so he got me involved in that, and then I started 00:11:51.130 --> 00:11:54.770 getting into biostatistics from there, so I was really like, kind of 00:11:54.770 --> 00:12:01.290 self -involved, like by opportunity I became biostatistics, you know? 00:12:01.290 --> 00:12:01.350 Thank you. 00:12:02.230 --> 00:12:02.890 Scott Berry: Mm-hmm. 00:12:03.100 --> 00:12:03.610 Okay. 00:12:03.990 --> 00:12:04.550 Byron Gajewski: Okay. 00:12:04.810 --> 00:12:07.900 Scott Berry: you're off at, uh, university of Kansas Medical Center, 00:12:08.290 --> 00:12:14.650 and I, I've been struck by the group you have there and really the innovative 00:12:14.650 --> 00:12:16.360 designs that are going on there. 00:12:16.720 --> 00:12:21.220 Now, I want to jump into one, the HOIT trial, but let's not go there yet. 00:12:21.700 --> 00:12:22.090 But 00:12:22.290 --> 00:12:22.790 Byron Gajewski: Sure. 00:12:22.990 --> 00:12:26.530 Scott Berry: bit about the group you have at KUMC, the 00:12:26.530 --> 00:12:28.720 innovative designs going on there. 00:12:28.720 --> 00:12:29.890 I think that would be great. 00:12:30.870 --> 00:12:34.550 Byron Gajewski: The group is, we have about 30 faculty members here. 00:12:35.690 --> 00:12:38.230 And we have several clinical trialists. 00:12:38.970 --> 00:12:41.530 Our department chair is a clinical trialist, meaning he 00:12:41.530 --> 00:12:44.950 does his research in clinical trials, does clinical trials. 00:12:45.830 --> 00:12:48.990 Uh, Matt Mayo, we have, uh, Joe Wick is a clinical 00:12:48.990 --> 00:12:50.050 trial, Bayesian adaptive. 00:12:50.190 --> 00:12:53.310 She's trained at Baylor as a Bayesian biostatistician 00:12:53.310 --> 00:12:54.310 at Baylor University. 00:12:54.310 --> 00:12:54.390 Okay. 00:12:55.245 --> 00:12:59.025 Um, Lynn Fadness is a, is a, is, is in our working group, and 00:12:59.025 --> 00:13:03.145 I'm probably forgetting, um, um, we have, we have, uh, some, some 00:13:03.145 --> 00:13:08.385 junior faculty, Kate Young, Lexi Brown, Bayesian, uh, statistician, 00:13:08.485 --> 00:13:11.725 some, some Bayesian, and we do a little bit of both, some of us, 00:13:11.805 --> 00:13:16.545 I do more Bayesian trials, in my end, my portfolio is more of that. 00:13:16.545 --> 00:13:23.065 And we have a working group of faculty, uh, staff, and students, 00:13:23.345 --> 00:13:25.040 and it's, uh, It's interesting. 00:13:25.260 --> 00:13:28.460 I'll, I'll, I'll say the name and then I'll give the acronym. 00:13:28.720 --> 00:13:32.080 It's, it's Fixed and Adaptive Clinical Trial 00:13:32.080 --> 00:13:33.340 Simulator Working Group. 00:13:33.420 --> 00:13:35.080 It's named after FACTS Software. 00:13:36.440 --> 00:13:39.880 And, um, in it, we kind of have a cool group. 00:13:39.940 --> 00:13:40.580 We mix things. 00:13:40.660 --> 00:13:42.520 We have this pre -specified rule. 00:13:42.660 --> 00:13:47.480 If there's a, if there's a webinar, a FACTS webinar during our working 00:13:47.480 --> 00:13:51.380 group, we will postpone a talk that we have internally to the 00:13:51.380 --> 00:13:52.800 later event and we'll do that. 00:13:53.180 --> 00:13:53.840 We'll go to that. 00:13:53.940 --> 00:13:56.280 And then we'll discuss it afterwards. 00:13:57.000 --> 00:14:02.200 But students and faculty, they do a novel clinical trial designs in it. 00:14:02.760 --> 00:14:04.640 Motivated from actual questions. 00:14:04.740 --> 00:14:05.780 And we've done this. 00:14:06.220 --> 00:14:08.560 There's a couple things we've done in that working group. 00:14:09.040 --> 00:14:10.440 That I think is kind of cool. 00:14:10.720 --> 00:14:11.460 We've done a trial. 00:14:12.460 --> 00:14:13.300 As a working group. 00:14:14.530 --> 00:14:15.340 Scott Berry: What does that mean? 00:14:15.370 --> 00:14:15.850 What does that 00:14:15.940 --> 00:14:19.720 Byron Gajewski: So what we did, Was to simplify things. 00:14:19.820 --> 00:14:24.825 To allow the students to experience designing and conducting a trial. 00:14:25.565 --> 00:14:29.625 We did a thing where I drive, I commute to work every day. 00:14:29.725 --> 00:14:33.305 I, myself, so it's a, it's a, it's a one of one clinical trial. 00:14:33.405 --> 00:14:34.365 It's a one of one trial. 00:14:34.465 --> 00:14:35.565 I shouldn't call it a clinical trial. 00:14:35.685 --> 00:14:36.965 It's a one of, one of one. 00:14:40.025 --> 00:14:42.585 And I have three different routes that I can take to work. 00:14:42.625 --> 00:14:44.025 And I have a route that I think works. 00:14:44.165 --> 00:14:46.925 So I have two experimental routes. 00:14:47.445 --> 00:14:51.565 And we did a response, we did a response adaptive randomization trial. 00:14:51.600 --> 00:14:53.200 They designed it. 00:14:53.920 --> 00:14:55.840 They looked at different operating characteristics. 00:14:56.020 --> 00:14:57.260 They looked at a fixed trial. 00:14:57.400 --> 00:14:58.300 They looked at adaptive. 00:14:58.480 --> 00:14:59.780 They picked an adaptive trial. 00:15:00.060 --> 00:15:03.660 They built the, uh, case report forms. 00:15:03.780 --> 00:15:05.760 They built the electronic data capturing. 00:15:06.160 --> 00:15:07.260 They did the randomization. 00:15:08.120 --> 00:15:09.780 Uh, they did that in RedCap. 00:15:10.840 --> 00:15:12.440 And, we implemented it. 00:15:12.460 --> 00:15:13.620 I went to an app in the morning. 00:15:13.700 --> 00:15:15.340 I went to my car, and I was randomized. 00:15:15.700 --> 00:15:17.220 And I did a, uh, A route. 00:15:17.460 --> 00:15:19.220 And then I input the data when I was done. 00:15:19.300 --> 00:15:20.540 I didn't do it during the drive. 00:15:21.175 --> 00:15:21.385 Scott Berry: Yeah. 00:15:21.690 --> 00:15:21.970 Yeah. 00:15:22.700 --> 00:15:25.380 Byron Gajewski: I, I started a clock and I went and I 00:15:25.380 --> 00:15:26.860 stopped the clock when I parked. 00:15:27.200 --> 00:15:27.960 We had protocols. 00:15:29.200 --> 00:15:33.440 And they discovered some, the trickeries and blinding things. 00:15:33.560 --> 00:15:34.600 Who needs to be blinded? 00:15:34.660 --> 00:15:35.780 Can you blind too much? 00:15:35.820 --> 00:15:36.620 Is that a risk? 00:15:36.940 --> 00:15:37.940 That showed up. 00:15:38.500 --> 00:15:43.720 Then they, they analyzed it per protocol and then they wrote up a 00:15:43.720 --> 00:15:46.260 paper and they've submitted the paper and it's under review right now. 00:15:46.795 --> 00:15:47.035 Scott Berry: Hmm. 00:15:48.120 --> 00:15:48.600 Byron Gajewski: Uh, 00:15:48.970 --> 00:15:53.185 Scott Berry: did the experimental routes end up quote unquote better? 00:15:56.740 --> 00:15:57.180 Byron Gajewski: yes. 00:15:57.400 --> 00:15:58.900 There was one route that was... 00:15:59.640 --> 00:16:03.940 Okay, so it ended up not hitting the formal trigger. 00:16:04.180 --> 00:16:07.560 So we would be able to FDA label it as the better route. 00:16:08.220 --> 00:16:11.160 We have like a posterior probability of 0 .94 00:16:11.160 --> 00:16:12.220 or something like that. 00:16:12.380 --> 00:16:13.260 That we have a better route. 00:16:14.125 --> 00:16:14.515 Scott Berry: Okay, 00:16:14.600 --> 00:16:15.080 Byron Gajewski: Yeah. 00:16:15.355 --> 00:16:19.285 Scott Berry: uh, I, I did a podcast with Lindsay, and you and I were talking 00:16:19.285 --> 00:16:23.035 about that ahead of time, where we talked about how to analyze ordinal endpoints 00:16:24.085 --> 00:16:27.865 is what's the way you analyze time? 00:16:27.865 --> 00:16:31.465 Is it mean, it linear? 00:16:31.735 --> 00:16:36.265 That 20 minutes is twice as bad as 10 minutes, or do you have some sort of 00:16:36.265 --> 00:16:39.265 loss function, minimax rule or something? 00:16:39.265 --> 00:16:40.825 It's just straight meantime. 00:16:41.380 --> 00:16:41.860 Byron Gajewski: Meantime. 00:16:42.720 --> 00:16:42.880 Yep. 00:16:43.120 --> 00:16:43.320 Yep. 00:16:43.560 --> 00:16:44.380 Normal stuff. 00:16:44.380 --> 00:16:45.500 Normal... 00:16:45.500 --> 00:16:47.800 Central Limit Theorem for the parameter estimates. 00:16:48.025 --> 00:16:48.505 Scott Berry: Okay. 00:16:48.775 --> 00:16:49.135 Okay. 00:16:49.195 --> 00:16:50.605 Nice, nice, 00:16:50.905 --> 00:16:53.725 Byron Gajewski: And then if you want to do, we talk about 00:16:53.725 --> 00:16:54.725 that actually with the group. 00:16:54.805 --> 00:16:58.825 We say if you want to do predictive district, you know, predictive times, 00:16:58.905 --> 00:17:00.105 it's not going to work very well. 00:17:01.085 --> 00:17:01.965 We need a different model. 00:17:02.480 --> 00:17:02.700 Scott Berry: yep. 00:17:03.205 --> 00:17:03.405 Byron Gajewski: Yeah. 00:17:04.855 --> 00:17:08.545 Scott Berry: Okay, so I, I, we want to jump into Hobe, but, um, some of 00:17:08.545 --> 00:17:15.565 the other trials that are not on you yourself as the, uh, as the, the patient, 00:17:16.045 --> 00:17:20.575 um, but you, you've done a number of Bayesian trials, the pain controls 00:17:20.575 --> 00:17:26.395 trials, the start trials, other trials, um, where you're also implementing them. 00:17:27.550 --> 00:17:32.050 You're creating red cap, you have a group there that sort of specializes in the 00:17:32.050 --> 00:17:37.120 operationalization of response, adaptive randomization, uh, and all of that, which 00:17:37.120 --> 00:17:38.935 has been a powerful part of these trials. 00:17:40.305 --> 00:17:41.025 Byron Gajewski: Yeah, absolutely. 00:17:41.665 --> 00:17:43.185 It's, it's, it's, it's been great. 00:17:43.285 --> 00:17:47.285 I mean, you mentioned, uh, uh, pain controls was interesting. 00:17:47.285 --> 00:17:47.705 Interesting. 00:17:48.475 --> 00:17:51.595 And it's, I tell this to Rick Barron, who's now at the 00:17:51.595 --> 00:17:54.215 University of Missouri, and I could talk the whole pot about 00:17:54.215 --> 00:17:57.175 the rivalry between KU and Mizzou. 00:17:57.995 --> 00:18:00.395 Um, maybe we don't want to go there, but it's 00:18:00.395 --> 00:18:01.335 an interesting rivalry. 00:18:02.175 --> 00:18:04.995 Um, so he's there now, and I saw him a couple weeks ago. 00:18:05.055 --> 00:18:06.535 Actually, my son goes to Mizzou. 00:18:07.095 --> 00:18:11.895 So, so I'm one of those KU guys that, that can go to 00:18:11.895 --> 00:18:14.575 Mizzou, wear Mizzou stuff, go to KU, wear KU stuff. 00:18:14.635 --> 00:18:15.615 I can go, you know. 00:18:15.690 --> 00:18:20.830 As long as they're not playing Texas A &M, that's right. 00:18:21.930 --> 00:18:24.870 Uh, but I told him, I, he was introducing me to some people out 00:18:24.870 --> 00:18:28.370 there and I said, I said, yeah, I work with Rick at, at KU and they 00:18:28.370 --> 00:18:31.490 do a, they do a thing with their head when they say you're at KU. 00:18:32.130 --> 00:18:33.110 It's what people do. 00:18:33.370 --> 00:18:36.170 But anyway, he, I, I said it's my favorite, it's 00:18:36.170 --> 00:18:37.210 one of my favorite trials. 00:18:37.710 --> 00:18:37.990 Pain control. 00:18:38.020 --> 00:18:42.040 Scott Berry: just so to, to give the one minute, uh, elevator pitch of 00:18:42.040 --> 00:18:43.540 what was the pain controls trial. 00:18:44.130 --> 00:18:48.730 Byron Gajewski: I describe for me my personal experience with 00:18:48.730 --> 00:18:49.610 pain controls and pain controls. 00:18:50.240 --> 00:18:52.520 is jump into the water to learn to swim. 00:18:54.040 --> 00:18:58.960 And what I mean by that is, um, it's interesting because, I 00:18:58.960 --> 00:19:01.240 don't know if you remember this, but Rick, I mentioned, mentioned 00:19:01.240 --> 00:19:02.800 Rick Barron, who's a neurologist. 00:19:03.360 --> 00:19:07.460 He went to a talk of yours, and, uh, I think it was at a 00:19:07.460 --> 00:19:10.620 translational, uh, clinical and translational research conference. 00:19:12.020 --> 00:19:15.900 And he liked what you had said about adaptive designs, about 00:19:15.900 --> 00:19:18.895 Bayesian adaptive designs, and he said, Hey, can we do that? 00:19:18.955 --> 00:19:20.235 Do you know anybody that can do that? 00:19:20.275 --> 00:19:23.035 And you said, well, Matt and Byron can do it at KU. 00:19:24.295 --> 00:19:27.555 And so then, because he and I, Matt and I had done some 00:19:27.555 --> 00:19:31.695 papers, some Bayesian papers about design and clinical trials, 00:19:32.375 --> 00:19:37.115 but we hadn't actually done any Bayesian or Bayesian adaptive trial. 00:19:37.195 --> 00:19:39.335 We hadn't actually formally designed them or implemented 00:19:39.335 --> 00:19:40.095 them at the time. 00:19:41.215 --> 00:19:42.935 And so you got me involved. 00:19:43.095 --> 00:19:44.375 You, you helped with that. 00:19:44.515 --> 00:19:45.755 We worked together on it. 00:19:45.835 --> 00:19:47.295 Melody worked on it from Barry. 00:19:47.855 --> 00:19:48.735 I remember that. 00:19:49.035 --> 00:19:51.255 We did a little bit of FACTS, but we did some R. 00:19:51.815 --> 00:19:53.615 And anyway, that's not the one minute thing. 00:19:53.795 --> 00:19:55.375 But the thing I want to tell you is I jumped 00:19:55.375 --> 00:19:56.875 right into it with that. 00:19:57.775 --> 00:20:03.155 We're talking, uh, uh, group sequential features, response 00:20:03.155 --> 00:20:06.295 adaptive randomization, identifying the maximum treatment. 00:20:06.455 --> 00:20:08.555 There's four arms in the trial, four treatments. 00:20:08.690 --> 00:20:10.510 There's two endpoints in it. 00:20:11.510 --> 00:20:13.250 And, uh, so it happened. 00:20:13.345 --> 00:20:14.485 Scott Berry: this, I, I think this 00:20:14.690 --> 00:20:15.290 Byron Gajewski: Yep. 00:20:16.435 --> 00:20:19.315 Scott Berry: comparative effectiveness of four different potentials for 00:20:19.375 --> 00:20:22.225 controlling pain, um, in, in this. 00:20:22.225 --> 00:20:22.555 Yep. 00:20:22.645 --> 00:20:22.795 Okay. 00:20:22.970 --> 00:20:23.110 Byron Gajewski: Yep. 00:20:23.490 --> 00:20:23.670 Yep. 00:20:23.910 --> 00:20:27.830 The disease was, or is, uh, cryptogenic neuropathy. 00:20:28.610 --> 00:20:32.090 Cryptogenic meaning they don't know why it caused the neuropathy. 00:20:32.290 --> 00:20:33.850 It was coined by Rick Barron. 00:20:33.950 --> 00:20:39.770 He discovered, or he defined that disease, uh, disease diagnosis. 00:20:41.535 --> 00:20:42.195 So 00:20:42.355 --> 00:20:45.325 Scott Berry: end result of the trial, which I don't remember. 00:20:45.675 --> 00:20:47.155 Byron Gajewski: there's four drugs. 00:20:47.355 --> 00:20:49.155 I'll be kind of abstract about it. 00:20:49.675 --> 00:20:50.495 Four drugs. 00:20:50.975 --> 00:20:53.655 We were trying to identify a single winner. 00:20:54.635 --> 00:20:59.615 We had put in the protocol at the 12th hour before, before finishing 00:20:59.615 --> 00:21:05.295 the protocol, um, that there be identification for loser arms. 00:21:05.635 --> 00:21:10.095 And there ended up being two, two loser arms, which, and so 00:21:10.095 --> 00:21:14.195 two are recommended for, Uh, diagnosed, or for, for treatment 00:21:14.195 --> 00:21:16.495 now in, in, in, in the realm. 00:21:17.055 --> 00:21:21.935 Um, and by the way, even the quote loser arms probably 00:21:21.935 --> 00:21:23.135 are better than do nothing. 00:21:23.635 --> 00:21:25.555 So, those are still good. 00:21:25.690 --> 00:21:26.680 Scott Berry: didn't have a do 00:21:26.775 --> 00:21:27.295 Byron Gajewski: We 00:21:27.400 --> 00:21:27.700 Scott Berry: though. 00:21:28.475 --> 00:21:29.735 Byron Gajewski: did not have the do nothing arm. 00:21:30.235 --> 00:21:31.395 No, no. 00:21:32.590 --> 00:21:33.010 Scott Berry: Okay. 00:21:33.010 --> 00:21:34.600 So let's, let's talk about hobo. 00:21:35.425 --> 00:21:40.405 Um, the, the HOIT trial, it's a, a, a very cool trial, adaptive 00:21:40.405 --> 00:21:43.825 trial and, uh, funded by the NIH. 00:21:43.825 --> 00:21:44.575 It's running. 00:21:44.845 --> 00:21:48.265 We are both blinded to the data, so we don't, we don't know the data, 00:21:48.505 --> 00:21:49.840 but tell me about the HOBIT trial. 00:21:51.395 --> 00:21:55.215 Byron Gajewski: Yeah, so, this trial that the disease is 00:21:55.215 --> 00:21:56.795 severe traumatic brain injury. 00:21:57.385 --> 00:22:01.525 So, think of a person gets in an accident, and it could 00:22:01.525 --> 00:22:04.985 be a car accident, a bike accident, it could be a fall, 00:22:06.125 --> 00:22:07.665 and their head gets hurt. 00:22:07.745 --> 00:22:09.925 They get hurt, they're in a coma, they're out. 00:22:11.005 --> 00:22:14.805 And that's usually diagnosed using a thing called the Glasgow Coma 00:22:14.805 --> 00:22:19.285 Scale, which you look at their eye response, their verbal response, 00:22:19.365 --> 00:22:21.325 motor responses, and it defines that. 00:22:21.325 --> 00:22:23.545 By the way, you kind of have to check to see if 00:22:23.545 --> 00:22:25.305 they're alcohol induced. 00:22:25.535 --> 00:22:30.175 Um, because if that happens, it's not a severe TBI. 00:22:30.395 --> 00:22:31.615 It's alcohol and it's temporary. 00:22:31.775 --> 00:22:33.395 It's not, it's not a long thing. 00:22:33.455 --> 00:22:36.235 So there's no good treatments for severe TBI. 00:22:37.535 --> 00:22:39.235 Um, we need to do better. 00:22:40.035 --> 00:22:46.755 And so, uh, Impetus of the, the, the HOBIT is, is hyperbaric 00:22:46.755 --> 00:22:50.175 oxygen treatment is, is, is looked at as being a potential 00:22:50.175 --> 00:22:51.935 treatment for severe TBI. 00:22:52.055 --> 00:22:55.285 It was studied preliminarily by Dr. 00:22:55.425 --> 00:22:59.165 Galen Roxworth out of Hennepin County Medical Center and the U. 00:22:59.225 --> 00:23:00.545 He's affiliated with the U. 00:23:01.030 --> 00:23:01.390 Scott Berry: Yep. 00:23:01.825 --> 00:23:03.705 Byron Gajewski: Um, and he has preliminary data. 00:23:03.785 --> 00:23:06.425 He's been studying this for decades. 00:23:07.225 --> 00:23:12.725 And it's become so dire that the NIH has said, you know, that is 00:23:12.725 --> 00:23:14.145 probably something we need to study. 00:23:14.245 --> 00:23:15.945 And the peer reviewers thought the same thing. 00:23:16.065 --> 00:23:17.425 You know, let's further study that. 00:23:18.085 --> 00:23:22.025 Um, the preliminary data suggests, so hyperbaric oxygen chamber is you, 00:23:22.115 --> 00:23:27.335 You put a patient into a, it's, it's indicated for, um, wounds and 00:23:27.335 --> 00:23:30.755 other medical conditions, and you put a, put a patient into a dive. 00:23:30.875 --> 00:23:35.035 They go into a, sometimes into a, what's called a monoplace chamber. 00:23:35.155 --> 00:23:36.115 It's a single chamber. 00:23:37.115 --> 00:23:41.155 Um, and there's a multiplace chamber where clinicians can go in with the 00:23:41.155 --> 00:23:42.575 patient and dive with the patient. 00:23:43.095 --> 00:23:44.155 So you're getting pressure. 00:23:44.875 --> 00:23:47.315 So you're getting better oxygen to the body. 00:23:48.795 --> 00:23:53.185 And Galen's theory is that, and And based on data, Uh, also 00:23:53.185 --> 00:23:58.925 empirically, empirically, uh, is that it reverses the ischemia 00:23:58.925 --> 00:24:04.045 in the brain, the damage in the brain, and can help with, uh, 00:24:04.105 --> 00:24:05.605 improving, improving their health. 00:24:06.065 --> 00:24:10.065 So he has preliminary data on that that suggests that there's 00:24:10.065 --> 00:24:12.785 a absolute improvement of 12 .7 00:24:12.785 --> 00:24:17.345 percent in favorable outcome between that and, and, and standard of care. 00:24:17.345 --> 00:24:19.085 Preliminary data. 00:24:19.540 --> 00:24:20.440 Scott Berry: Okay, but the trial is 00:24:20.555 --> 00:24:21.235 Byron Gajewski: Yes, 00:24:21.550 --> 00:24:25.420 Scott Berry: a two arm trial of hyperbaric oxygen against none. 00:24:25.690 --> 00:24:29.350 Uh, it's a, it's a more complex treatment, and so you have 00:24:29.350 --> 00:24:31.075 multiple experimental arms. 00:24:32.535 --> 00:24:32.835 Byron Gajewski: yes. 00:24:32.955 --> 00:24:36.975 As I, as I mentioned, there's a dive, but there's, so, so 00:24:36.975 --> 00:24:40.275 atmospheric pressure is a 1 .0 00:24:40.275 --> 00:24:41.415 at sea level. 00:24:42.495 --> 00:24:44.335 And so 1 .0 00:24:44.335 --> 00:24:47.395 pressure can be increased to 1 .5, 00:24:48.055 --> 00:24:50.655 to 2, to 2 .5, 00:24:50.695 --> 00:24:52.695 and the patient goes into a dive. 00:24:52.695 --> 00:24:56.095 So it's like, it's like a diver going in the ocean, going 00:24:56.095 --> 00:24:58.295 under the, under the water, or if you're an airplane. 00:24:58.570 --> 00:25:02.150 That pressure change happens and it happens slowly so 00:25:02.150 --> 00:25:06.410 you're not hurting the, the, the patient or the clinician 00:25:06.410 --> 00:25:08.330 who's in a monoplace chamber. 00:25:09.490 --> 00:25:13.770 And so we're looking at, we'll cast a wide net on this. 00:25:14.410 --> 00:25:18.830 We're in a kind of a phase 2 case, phase 2B if you will. 00:25:19.710 --> 00:25:24.465 And, We have three different levels of hyperbaric oxygen, and then 00:25:24.465 --> 00:25:27.825 there's a thing called normobaric hypoxia, which is not pressure 00:25:27.825 --> 00:25:29.065 -reduced, it's just oxygen. 00:25:29.845 --> 00:25:33.925 And we have that in addition to it, so that turns into six arms, and 00:25:33.925 --> 00:25:35.705 then we have normobaric by itself. 00:25:35.765 --> 00:25:37.845 You can kind of think of the arms as being eight 00:25:37.845 --> 00:25:39.785 arms, kind of a four by two. 00:25:40.575 --> 00:25:40.905 Scott Berry: Yep. 00:25:41.595 --> 00:25:41.775 Yep. 00:25:42.005 --> 00:25:46.205 Byron Gajewski: Four hyperbaric oxygen levels and two normobaric hypoxia. 00:25:46.965 --> 00:25:48.945 And then the one with 1 .0 00:25:48.945 --> 00:25:53.165 ETA and non -normabaric hypoxia, that's the control, that's 00:25:53.165 --> 00:25:57.365 the standard of, of care, which, you know, um, yeah. 00:25:57.865 --> 00:25:58.765 Scott Berry: it wa it's, it's a 00:25:59.065 --> 00:25:59.585 Byron Gajewski: Not 00:25:59.635 --> 00:26:01.465 Scott Berry: So they go through the procedure. 00:26:01.465 --> 00:26:05.965 They're just not getting the, the, the mechanism of it. 00:26:06.665 --> 00:26:08.905 Byron Gajewski: really, they do not go, they do not go into, 00:26:09.325 --> 00:26:13.465 no, no, no, they're just treated standard. 00:26:14.035 --> 00:26:14.365 Scott Berry: Oh, so 00:26:14.465 --> 00:26:15.285 Byron Gajewski: No. 00:26:15.475 --> 00:26:16.675 Scott Berry: do a fake dive. 00:26:16.735 --> 00:26:16.915 Okay. 00:26:20.485 --> 00:26:24.895 Okay, so you have these multiple arms. 00:26:25.165 --> 00:26:29.245 What's the, uh, and the endpoint in the trial is what are 00:26:29.325 --> 00:26:34.345 Byron Gajewski: a, yeah, it's a scale called Glasgow Outcome Scale Extended. 00:26:34.430 --> 00:26:38.430 That's six months, and it, it's a, it's an ordinal scale 00:26:38.430 --> 00:26:39.770 that goes from one to eight. 00:26:40.930 --> 00:26:45.390 And one is good, eight is bad, one is, I'm sorry, 00:26:45.470 --> 00:26:46.030 it's the other way around. 00:26:46.310 --> 00:26:48.250 One is, one is bad, eight is good. 00:26:49.270 --> 00:26:54.170 Um, so death is a one, vegetative state is a two. 00:26:54.310 --> 00:26:57.490 Kind of like the modified Rankine scale, a lot of people 00:26:57.490 --> 00:26:58.550 treat those as the same. 00:26:58.990 --> 00:27:01.610 The one and two, the, the death, the vegetative state. 00:27:01.610 --> 00:27:01.670 Thank you. 00:27:01.715 --> 00:27:06.735 Three through six is a kind of a fine, different gradations 00:27:06.735 --> 00:27:08.295 of moderate disability. 00:27:09.595 --> 00:27:14.055 And then a seven to eight is good recovery, gradation, good recovery. 00:27:14.275 --> 00:27:15.575 So an eight is really good. 00:27:16.935 --> 00:27:19.435 Um, and then we, it's a slide, it's a sliding dichotomy. 00:27:19.635 --> 00:27:24.295 So depending on the severity of the patient in the beginning at baseline, 00:27:24.435 --> 00:27:28.695 you have different definitions of success and it's a binary cutoff. 00:27:28.775 --> 00:27:30.815 This is a, this is a binary. 00:27:31.325 --> 00:27:34.425 Uh, outcome that we analyze at the end of the day. 00:27:34.905 --> 00:27:35.845 Um, uh, we, 00:27:36.785 --> 00:27:37.965 that's 00:27:38.215 --> 00:27:41.575 Scott Berry: on where, what your baseline status is, defines 00:27:41.575 --> 00:27:43.345 differently what a response is. 00:27:43.845 --> 00:27:44.025 Byron Gajewski: right. 00:27:44.045 --> 00:27:45.485 Scott Berry: sliding dichotomy part of 00:27:45.545 --> 00:27:45.945 Byron Gajewski: Mm -hmm, 00:27:46.475 --> 00:27:49.205 Scott Berry: And now the design is adaptive. 00:27:49.615 --> 00:27:52.715 So you have these multiple arms in the trial. 00:27:52.715 --> 00:27:56.765 You have a control arm and the endpoint's at six months. 00:27:56.765 --> 00:27:59.070 But what, what are the adaptive parts of the design? 00:27:59.845 --> 00:28:00.745 Byron Gajewski: sure, sure. 00:28:00.825 --> 00:28:05.705 There are, there are several adaptive features to the design. 00:28:06.345 --> 00:28:09.005 Um, we use response adaptive randomization. 00:28:09.775 --> 00:28:12.615 And we do it on the primary endpoint. 00:28:13.015 --> 00:28:17.335 We do a information formula, which combines the probability. 00:28:17.875 --> 00:28:20.255 So, so by the way, we're trying to say the, the, the 00:28:20.255 --> 00:28:24.495 probability that the active arm is best among all active arms. 00:28:25.105 --> 00:28:25.525 Mm-hmm. 00:28:25.525 --> 00:28:28.655 Uh, and we, we put variance and sample size of that, 00:28:28.735 --> 00:28:29.795 uh, in that arm already. 00:28:30.175 --> 00:28:32.015 And, and, and do that. 00:28:32.115 --> 00:28:33.315 We do an information formula. 00:28:33.495 --> 00:28:37.195 So it uses both the best arm as well as the information in 00:28:37.195 --> 00:28:38.435 that arm in a balanced way. 00:28:38.460 --> 00:28:43.140 Um, and then we also have, like, group sequential features to it. 00:28:43.200 --> 00:28:49.680 So we, we do RAR every so often, and it's, it's 00:28:49.680 --> 00:28:52.060 precisely the enrollment times. 00:28:52.280 --> 00:28:54.200 Actually, it's not just every so often. 00:28:54.460 --> 00:28:59.000 It's very specific time or very specific enrollment points. 00:28:59.660 --> 00:29:04.260 And then we do, uh, we test for whether we've achieved 00:29:04.260 --> 00:29:07.120 success or futility at interims. 00:29:07.245 --> 00:29:14.525 We have a maximum sample size of 200, and 20 of the, of the 00:29:14.525 --> 00:29:18.965 participants have been randomized to the control throughout the trial. 00:29:19.890 --> 00:29:20.280 Scott Berry: So the, Hmm. 00:29:20.280 --> 00:29:23.815 So the RAR changes the active arms 00:29:23.825 --> 00:29:24.085 Byron Gajewski: Uh -huh. 00:29:24.085 --> 00:29:28.525 Scott Berry: on pressure and, uh, this, this NBH. 00:29:28.765 --> 00:29:29.365 Uh, 00:29:29.805 --> 00:29:30.065 Byron Gajewski: Yep. 00:29:30.705 --> 00:29:30.905 Yep. 00:29:31.445 --> 00:29:31.985 That's 00:29:32.245 --> 00:29:35.305 Scott Berry: the trial could stop for futility approximately every 00:29:35.305 --> 00:29:37.075 20 patients or something like that. 00:29:37.075 --> 00:29:37.195 You 00:29:37.265 --> 00:29:37.445 Byron Gajewski: right. 00:29:38.265 --> 00:29:38.465 Yep. 00:29:38.745 --> 00:29:39.825 We put 20 patients. 00:29:39.965 --> 00:29:41.485 It kicked in at 116. 00:29:41.485 --> 00:29:47.125 You go 116, 136, 156, 176. 00:29:47.235 --> 00:29:48.955 And then 200 is the maximum. 00:29:49.055 --> 00:29:50.935 I'm talking about enrolled here. 00:29:51.575 --> 00:29:57.675 So the 116th, uh, patient that was enrolled to be a participant, we 00:29:57.675 --> 00:29:59.035 started, we did an interim analysis. 00:29:59.095 --> 00:30:03.435 And I say we, uh, Jonathan and Renee did the interim analysis. 00:30:03.925 --> 00:30:05.185 Scott Berry: Yeah, so this is, uh, 00:30:05.235 --> 00:30:05.795 Byron Gajewski: That's 00:30:06.175 --> 00:30:07.225 Scott Berry: of South Carolina. 00:30:07.375 --> 00:30:07.515 Byron Gajewski: right. 00:30:07.600 --> 00:30:08.000 That's 00:30:08.245 --> 00:30:11.035 Scott Berry: coordinatings, they're implementing the adaptive design. 00:30:11.540 --> 00:30:11.680 Byron Gajewski: right. 00:30:11.815 --> 00:30:12.145 Scott Berry: yep. 00:30:12.235 --> 00:30:12.385 Yep. 00:30:12.415 --> 00:30:12.745 Okay. 00:30:12.955 --> 00:30:13.255 Yep. 00:30:15.490 --> 00:30:20.440 Alright, so, uh, and, and do you know roughly the, this trial's enrolling 00:30:20.440 --> 00:30:23.980 you're, you're roughly at a sample size of 00:30:25.100 --> 00:30:28.000 Byron Gajewski: Yeah, so we're, we, we are at 159. 00:30:28.680 --> 00:30:31.440 So the latest interim we did was 156. 00:30:32.340 --> 00:30:39.600 And I know, so we talked about this a little bit, uh, I, I think 00:30:39.600 --> 00:30:44.960 of, so I'm, I'm, as you mentioned, I'm a blinded, I'm a blinded STEM. 00:30:45.010 --> 00:30:46.450 Scott Berry: You're a blinded advisor. 00:30:46.510 --> 00:30:46.750 To? 00:30:46.750 --> 00:30:47.440 To, to, 00:30:47.640 --> 00:30:48.820 Byron Gajewski: Right, right. 00:30:48.920 --> 00:30:51.760 And so I don't know what's going on. 00:30:52.365 --> 00:30:55.565 Except I, I, I use the analogy of poker. 00:30:56.325 --> 00:30:57.585 I'm playing poker here. 00:30:57.865 --> 00:30:59.625 And obviously we're not gambling with Hobbit. 00:30:59.705 --> 00:31:01.425 We're not, that's not what I'm trying to say. 00:31:01.885 --> 00:31:03.425 I'm just trying to use an analogy. 00:31:03.595 --> 00:31:03.885 Scott Berry: yeah. 00:31:05.745 --> 00:31:09.385 Byron Gajewski: Um, where I'm playing against somebody, and I 00:31:09.385 --> 00:31:10.665 don't know what their hand is. 00:31:11.665 --> 00:31:12.825 But I have some telltales. 00:31:12.945 --> 00:31:13.725 I know how they're betting. 00:31:14.625 --> 00:31:17.885 I might know that they're hitting their ear or, or, or nose. 00:31:18.145 --> 00:31:19.525 I don't know if I believe that. 00:31:19.525 --> 00:31:19.885 But, 00:31:20.485 --> 00:31:20.905 um. 00:31:21.925 --> 00:31:26.165 So I have some tell, the telltale I have is that 156, we have 00:31:26.165 --> 00:31:28.005 not achieved success or futility. 00:31:28.885 --> 00:31:31.565 I have some information about the trial. 00:31:31.705 --> 00:31:35.725 I know that it hasn't, it hasn't hit those workers. 00:31:35.755 --> 00:31:38.665 Scott Berry: And you don't know the randomization to the arms and all of 00:31:38.665 --> 00:31:41.965 this, but this is, I mean, in some ways this is actually a fantastic trial. 00:31:41.965 --> 00:31:46.255 This is a, uh, uh, this is a syndrome where. 00:31:46.585 --> 00:31:50.485 Nothing works is severe concussion, severe traumatic brain 00:31:50.485 --> 00:31:52.825 injury, many different drugs. 00:31:52.945 --> 00:31:57.715 Now, there's actually a brain adaptive platform trial that's running in Canada. 00:31:57.715 --> 00:32:02.035 They're investigating multiple things, uh, but largely this is a really 00:32:02.035 --> 00:32:04.225 refractory disease to treatment. 00:32:04.705 --> 00:32:10.675 So the, the ability for this to show and explore multiple ways to give hyperbaric 00:32:10.675 --> 00:32:15.505 oxygen in one trial and demonstrate lump some level of efficacy would be. 00:32:16.255 --> 00:32:19.555 Complete game changing, uh, with positivity. 00:32:19.555 --> 00:32:22.525 Now we, we don't know that, you know, this could be in completely 00:32:22.525 --> 00:32:26.125 ineffective, uh, we don't know the data, but the design is a really, 00:32:26.125 --> 00:32:28.555 really cool design, uh, in this setting. 00:32:28.555 --> 00:32:33.415 So, uh, really looking forward to seeing how the randomization came out. 00:32:33.415 --> 00:32:35.845 The trial came out fantastic. 00:32:36.585 --> 00:32:36.805 Byron Gajewski: yeah. 00:32:37.145 --> 00:32:37.665 So, 00:32:38.845 --> 00:32:41.605 Scott Berry: Uh, and now when you are working with. 00:32:41.965 --> 00:32:46.975 The very ac various academic groups and good receptiveness 00:32:46.975 --> 00:32:49.315 to Bayesian adaptive designs. 00:32:49.315 --> 00:32:54.265 You feel like this is, uh, hard discussions, easy discussions. 00:32:55.545 --> 00:32:57.365 Byron Gajewski: um, easy. 00:32:58.125 --> 00:32:59.525 It's really remarkable. 00:33:00.165 --> 00:33:02.025 Um, I'm really surprised. 00:33:02.170 --> 00:33:05.430 I really thought in the beginning it would be a real struggle because I 00:33:05.430 --> 00:33:11.190 was kind of brought up, you know, in the 90s, we were still in this, 00:33:11.410 --> 00:33:15.330 when I was at Texas A &M, we were still in this time of great debate. 00:33:16.390 --> 00:33:19.690 Frequentists in Beijing used to struggle, but the debate centered 00:33:19.690 --> 00:33:24.750 around kind of mathematical, philosophical struggles. 00:33:26.170 --> 00:33:31.430 Um, you know, you talk about philosophies, what is science, and 00:33:31.560 --> 00:33:38.180 When you talk to collaborators who are non -statistical, and you say, 00:33:38.580 --> 00:33:43.020 Hey, how would you like to interpret your results as the probability 00:33:43.020 --> 00:33:46.420 of a treatment effect, given data? 00:33:46.940 --> 00:33:47.880 Well, that sounds great. 00:33:48.695 --> 00:33:48.895 Scott Berry: Yeah. 00:33:49.080 --> 00:33:50.100 Byron Gajewski: Well, we'll just go from there. 00:33:50.485 --> 00:33:50.775 Scott Berry: Yeah. 00:33:51.120 --> 00:33:52.960 Byron Gajewski: And, and people really like it. 00:33:53.000 --> 00:33:56.580 They like, also, that I really, and I don't think this is necessarily 00:33:56.580 --> 00:34:00.440 a Bayesian thing, but I think it's easier to sell in the Bayesian. 00:34:00.555 --> 00:34:04.515 It is, and I think you talked about this on a pod, about reading 00:34:04.515 --> 00:34:08.615 the JAMA article about labels, and I alluded to it earlier. 00:34:10.695 --> 00:34:17.295 You know, yes or no, this binary world in science and medicine 00:34:17.295 --> 00:34:19.395 is very unappealing to me. 00:34:19.455 --> 00:34:22.255 I like to think of science as a spectrum. 00:34:22.255 --> 00:34:26.055 And I think when you can say the probability of a 00:34:26.055 --> 00:34:28.155 treatment effect is .96, 00:34:28.495 --> 00:34:29.715 I think it was Tesla. 00:34:31.010 --> 00:34:32.170 Trial and stroke. 00:34:32.730 --> 00:34:36.230 I think the trigger was, I want to say it was 0.975 00:34:36.390 --> 00:34:40.450 or something for the, and I think it hit 0.974 00:34:40.450 --> 00:34:40.930 or something. 00:34:41.750 --> 00:34:44.880 And then, New England Journal of Medicine, I think said, well, 00:34:44.990 --> 00:34:46.890 it doesn't, it didn't, they didn't say it doesn't work. 00:34:47.010 --> 00:34:51.070 They just said, you couldn't say that it worked, that endovascular 00:34:51.070 --> 00:34:54.870 therapy was better than, and it's really kind of like, I, anyway. 00:34:55.270 --> 00:34:57.670 I think people, I talk to pediatricians, we 00:34:57.670 --> 00:34:58.710 have a network here. 00:34:58.855 --> 00:35:04.535 Uh, that, that Matt Mayo got a, uh, data coordinating center for, 00:35:04.615 --> 00:35:09.735 it's called DCOC, and it's a rural pediatric network that he's in charge 00:35:09.735 --> 00:35:13.295 of as a statistician, the Data Coordinating Center and Operations. 00:35:14.855 --> 00:35:19.975 And I talked about this concept, and they were like, oh, oh, but I'm 00:35:19.975 --> 00:35:23.715 talking about like, maybe I don't want to say names necessarily, but, 00:35:24.825 --> 00:35:27.625 Scientists in the room, we were talking, and they just said, Yeah, 00:35:27.765 --> 00:35:31.365 that makes a lot, this concept of labeling, not labeling it yes or no, 00:35:31.445 --> 00:35:36.085 but labeling the probability of, of a hypothesis, they really liked it. 00:35:36.125 --> 00:35:36.605 So I guess, 00:35:39.225 --> 00:35:40.025 it's been easy. 00:35:40.085 --> 00:35:41.365 It's been an easy sell, I would say. 00:35:41.565 --> 00:35:43.985 And I don't mean to say a sell. 00:35:44.085 --> 00:35:47.605 I just mean to say, I like to work with investigators and have 00:35:47.605 --> 00:35:49.705 them, let them make decisions. 00:35:49.705 --> 00:35:53.045 I just present kind of ideas. 00:35:53.845 --> 00:35:54.055 Scott Berry: Yeah. 00:35:54.580 --> 00:35:55.180 Byron Gajewski: That's 00:35:55.345 --> 00:35:58.795 Scott Berry: this trial that you're running, the HOIT trial that you're 00:35:58.795 --> 00:36:02.875 running, the primary analysis is going to be, what's the probability 00:36:02.875 --> 00:36:05.575 that this arm is better than control? 00:36:05.960 --> 00:36:06.160 Byron Gajewski: right. 00:36:06.655 --> 00:36:09.895 Scott Berry: I believe, quote unquote, success in the trial is 00:36:09.895 --> 00:36:12.175 greater than an 85% probability. 00:36:12.355 --> 00:36:12.895 It's not. 00:36:13.660 --> 00:36:18.190 An FDA approval threshold of 97.5 00:36:18.190 --> 00:36:19.030 or 99. 00:36:19.390 --> 00:36:24.280 And so the really, the interpretation of this is a, is a quantitative scale. 00:36:24.490 --> 00:36:27.580 What's the probability that this is beneficial? 00:36:27.850 --> 00:36:34.000 And it may be that 90% is practice changing, maybe 50%. 00:36:34.000 --> 00:36:37.330 You know, I mean, so it, it, it, it's the idea that. 00:36:38.200 --> 00:36:42.280 Running this trial, and this is a hard trial to run. 00:36:42.280 --> 00:36:44.320 It's been long, it's been slow. 00:36:44.500 --> 00:36:48.310 The idea that there's one answer that comes out, yes or no, seems absurd. 00:36:48.520 --> 00:36:48.820 Byron Gajewski: Yeah. 00:36:48.910 --> 00:36:51.160 Scott Berry: And the Bayesian analysis is gonna help 00:36:51.160 --> 00:36:52.810 quantify, given the information. 00:36:52.810 --> 00:36:56.140 You do have a dose response model across these arms. 00:36:56.140 --> 00:36:58.120 So it's not like multiplicities of this. 00:36:58.120 --> 00:37:01.780 But given the entirety of the data and the modeling, what's the probability 00:37:01.780 --> 00:37:05.710 that this is, uh, beneficial, which is, which is really cool. 00:37:06.200 --> 00:37:09.540 Byron Gajewski: Oh, and by the way, um, there's also an additional 00:37:09.540 --> 00:37:11.120 success criteria for that. 00:37:11.260 --> 00:37:14.340 Not only is it the probability has to be bigger than 0 .85, 00:37:15.180 --> 00:37:19.360 but there's an and, we predict, posterior predictive 00:37:19.360 --> 00:37:21.240 probability of phase 3 success. 00:37:21.630 --> 00:37:21.850 Scott Berry: Hmm. 00:37:22.660 --> 00:37:25.120 Byron Gajewski: So the future phase 3 would be a 00:37:25.120 --> 00:37:28.180 thousand participants, 0 .025 00:37:28.180 --> 00:37:30.490 type 1 error on one -sided, and 0 .025 00:37:30.490 --> 00:37:30.600 type 2 error on one -sided. 00:37:30.700 --> 00:37:34.380 Chi -square on the endpoint, one -to -one randomization. 00:37:35.100 --> 00:37:37.960 If that probability of success is bigger than 0 .5, 00:37:38.120 --> 00:37:40.240 then that, so it's an and statement. 00:37:40.660 --> 00:37:42.010 Scott Berry: Yeah, so th th this 00:37:42.140 --> 00:37:42.700 Byron Gajewski: Right, 00:37:43.480 --> 00:37:47.320 Scott Berry: proof of concept phase two trial, and if it's, if it 00:37:47.320 --> 00:37:52.660 predicts that phase three would be reasonable likelihood of success, that 00:37:52.660 --> 00:37:55.390 that's a, a positive of this trial. 00:37:56.380 --> 00:37:57.500 Byron Gajewski: right, right. 00:37:57.700 --> 00:37:58.090 Scott Berry: cool. 00:37:59.320 --> 00:38:00.130 Very cool. 00:38:01.570 --> 00:38:06.250 Alright, so, uh, lots of, uh, fun Bayesian trials going on. 00:38:06.250 --> 00:38:09.760 Any other exciting things going on at KUMC? 00:38:10.100 --> 00:38:13.980 Byron Gajewski: Um, no, I mean, I, I think that since, since pain 00:38:13.980 --> 00:38:18.300 controls, I really, I've been, 00:38:18.640 --> 00:38:20.060 so 00:38:20.110 --> 00:38:22.360 Scott Berry: So you've, I know you've had a number of students that are 00:38:22.360 --> 00:38:27.460 now working in this area of Bayesian trials, um, uh, innovative trial design. 00:38:27.790 --> 00:38:28.750 Is that. 00:38:28.795 --> 00:38:34.435 Uh, find that students find that an exciting area to go into? 00:38:35.740 --> 00:38:38.160 Byron Gajewski: the answer, short answer is absolutely. 00:38:38.160 --> 00:38:38.260 Right. 00:38:38.305 --> 00:38:43.145 You, you actually, it's, it's kind of cool. 00:38:44.125 --> 00:38:46.765 They, they're excited to do it. 00:38:46.885 --> 00:38:50.045 They're, they're, you know, I actually, speaking of Melanie, I 00:38:50.045 --> 00:38:53.325 had this discussion with Melanie years ago at Joint Statistics. 00:38:53.550 --> 00:38:55.405 Scott Berry: Cantana, who's at Berry Consultants. 00:38:55.405 --> 00:38:55.525 Yep. 00:38:56.225 --> 00:38:59.085 Byron Gajewski: I had this discussion with her about building 00:38:59.085 --> 00:39:02.405 adaptive designs and how it's like being a kid in a sandbox. 00:39:02.405 --> 00:39:04.165 You're just building it. 00:39:04.310 --> 00:39:08.050 you know, a sand castle or legos or something. 00:39:08.210 --> 00:39:10.270 It's very rewarding and fun, and I think the 00:39:10.270 --> 00:39:11.570 students think the same way. 00:39:12.070 --> 00:39:14.630 They think the same, and it kind of is reassuring to know 00:39:14.630 --> 00:39:17.170 that other people appreciate it too, and I'm not just some 00:39:20.010 --> 00:39:22.870 strange guy who loves building adaptive designs. 00:39:23.110 --> 00:39:24.070 Scott Berry: Yeah, yeah. 00:39:24.310 --> 00:39:28.150 Uh, that I, I, I don't know if that's a fear that we as statisticians 00:39:28.150 --> 00:39:31.540 generally have that we find really, really cool things and wonder 00:39:31.720 --> 00:39:33.130 does anybody else think this? 00:39:33.130 --> 00:39:33.790 Is that cool? 00:39:33.790 --> 00:39:35.350 This is, this is super cool. 00:39:35.410 --> 00:39:35.710 Yeah. 00:39:36.470 --> 00:39:36.970 Byron Gajewski: Thanks 00:39:37.165 --> 00:39:37.495 Scott Berry: Yeah. 00:39:37.795 --> 00:39:38.665 Oh, that's awesome. 00:39:38.935 --> 00:39:43.795 Uh, the, the, you're, you're, you're turning out, uh, uh, new design students. 00:39:43.795 --> 00:39:44.935 That's, that's fabulous. 00:39:44.935 --> 00:39:47.785 The work going on at KUMC is really exciting. 00:39:48.175 --> 00:39:49.645 Look for the hobo trial. 00:39:49.855 --> 00:39:51.835 Um, it's not over. 00:39:51.835 --> 00:39:52.795 It's still running. 00:39:53.035 --> 00:39:58.120 Uh, but Byron, thanks for joining us, uh, here in the interim. 00:39:59.850 --> 00:40:00.810 Byron Gajewski: for having me. 00:40:00.970 --> 00:40:02.870 I really enjoyed the discussion, Scott. 00:40:04.390 --> 00:40:06.670 Scott Berry: And everybody, thanks for joining us. 00:40:06.670 --> 00:40:07.930 Thanks for tuning in. 00:40:08.260 --> 00:40:11.380 Until next time, we will be here in the interim.