WEBVTT
1
00:00:00.100 --> 00:00:18.040
[upbeat music] Welcome back to another episode of Vanguard Radio. I'm your host, J. Richard Jones. It's a pleasure to be with you today, and we're gonna get into some very interesting topics related to healthcare.
2
00:00:18.500 --> 00:00:27.460
I have two very special guests that are joining me, and they have a very progressive company that is incorporating healthcare and technology.
3
00:00:27.920 --> 00:00:35.560
And just before the show, we were chatting a little bit about this, and parts of the discussion were touching around the role of AI.
4
00:00:35.800 --> 00:00:49.040
There's an evolution going on in terms of training, and it's setting an incredibly fast pace in terms of the benefits of being able to improve healthcare training and accessibility.
5
00:00:49.080 --> 00:01:01.060
The other side of it, though, is with technology, the benefit is you're able to take training and education for medical professionals and move it from the class into training at scale.
6
00:01:01.500 --> 00:01:13.400
This provides opportunity for better care and for remote training right across the board. Our guests today joining us are Mr. Raja Khanna, who is the CEO of Lumeto, and Mr.
7
00:01:13.920 --> 00:01:26.440
Kavi Maharaj, who is the chief product officer or CPO with Lumeto. Now, if you're not familiar with Lumeto, they were just recently at our simulation and training forum put on by Vanguard.
8
00:01:27.160 --> 00:01:42.320
They are a Canadian healthcare training and dual-use simulation company whose platform involve XR, and it delivers clinically proven AI-enabled and customizable immersive training for healthcare and defense.
9
00:01:42.750 --> 00:01:55.340
It's pretty neat stuff. It was amazing to see their demonstrations, and they'll probably be at CANSEC this year, too, if you wanna meet with them. So gentlemen, welcome to the show. It's a pleasure to be here.
10
00:01:55.920 --> 00:02:06.120
Thank you for having us. It's great to have you on. And generally, we're gonna talk a little bit about the training, training the future force, really, and the advantages of AI and scalable simulation for readiness.
11
00:02:06.660 --> 00:02:16.040
Raja, I'm gonna jump right in. Lumeto is known as a healthcare simulation company. How should the defense community be thinking about who you are?
12
00:02:18.000 --> 00:02:27.150
Yeah, we've been around for about six years training healthcare professionals in the civilian sense in hospitals and nursing schools and medical schools.
13
00:02:27.900 --> 00:02:42.020
And really at our core, we're using AI to simulate complex medical situations to give the chance for clinicians to basically practice at scale without the need for mannequins or physically being somewhere.
14
00:02:42.590 --> 00:02:47.210
That same paradigm, it applies directly to the defense sector.
15
00:02:47.820 --> 00:03:07.100
Uh, we've been lucky enough to do a few, uh, pilots with the Canadian Armed Forces, and, uh, we see big opportunity here because when you have a distributed, you know, uh, force that all need to have standardized set of skills and all need a chance to practice and gain confidence in complex scenarios that are hard to learn in a classroom, how do you do that?
16
00:03:07.520 --> 00:03:23.380
Uh, we've come up with a solution that is, uh, really about immersive simulation that uses AI to basically mimic these types of environments and the stress and complexity of those types of environments, giving people a chance to practice, learn, build confidence, and build, uh, uh, medical skill.
17
00:03:23.680 --> 00:03:32.560
So that's who we are. We're pretty excited about the dual use approach for our business, taking our, what we've done in the hospital and academic world and bringing it to defense.
18
00:03:34.360 --> 00:03:41.000
How's it been received so far by the defense community, Raja? Uh, it's been overwhelmingly positive.
19
00:03:41.180 --> 00:03:47.460
Uh, I think our use cases and our thesis for our business, we're, we're working in hospitals and academic institutions around the world now.
20
00:03:48.040 --> 00:04:00.900
It fits even better in defense because here you have people that need to be trained that are almost always spread out across, uh, the world, if not a nation. Uh, and this technology works incredibly effectively remotely.
21
00:04:00.920 --> 00:04:13.040
You no longer have to fly people into training centers. It's easier to deploy at scale. Uh, there's more opportunity for training, uh, when you have this type of digital technology that can be adopted anywhere.
22
00:04:13.080 --> 00:04:16.660
You can have an instructor in one city and, you know, learners in another city.
23
00:04:17.220 --> 00:04:34.700
Uh, that's incredibly powerful, and I think just AI unlocks all kinds of new modalities that even if you did fly them all to the same place and train them on a mannequin, which is how it often happens today, that mannequin usually can't talk, or you have to hire actors as standardized patients.
24
00:04:34.820 --> 00:04:48.840
It has been traditionally very difficult to train complex medical scenarios. Now we can create those like a mass casualty event in a simulated AI-driven virtual environment, uh, and it's, uh, pretty compelling.
25
00:04:48.969 --> 00:04:58.420
So I'd say the response has been overwhelmingly positive. Uh, I think the defense community have already adopted these types of modality for other combat training and so on.
26
00:04:58.770 --> 00:05:06.320
And so in healthcare, I think in particular, this is the next, uh, frontier, and, and people are excited to engage. I'm not surprised.
27
00:05:06.400 --> 00:05:18.010
I mean, the technology that is allowing for this type of training and access to go on is just incredible, and remote access, I mean, just imagine all of the different things that can be done.
28
00:05:18.660 --> 00:05:30.620
Kavi, I wanna come over to you now. The term simulation means very different things to people in the defense world. How do you position what Lumeto does with that landscape?
29
00:05:32.240 --> 00:05:34.840
Yeah, that's a great question, and it's an important distinction.
30
00:05:34.920 --> 00:05:51.060
I'll tell you, we first of all fit into synthetic simulation, so we of course know that the military and most training, uh, modalities for simulation today are what's called high fidelity, so that's realistic training with mannequins or sometimes real people, non-scalable.
31
00:05:51.720 --> 00:05:57.860
And within synthetic simulation, there are really two categories that we see happening, uh, and existing today.
32
00:05:58.040 --> 00:06:10.562
One of them is kinetic-based simulation, a lot of familiarity with that, mostly weapon systems and vehicle operation, tactical type, uh, maneuvers. And training programs around that.
33
00:06:10.852 --> 00:06:34.892
We are actually focused on human factor training, and that difference there is mainly, uh, focused around human performance, communication, team coordination, clinical decision-making under stress, and overall just trying to get a sense of confidence in actually running the simulation in a firsthand, first-person experience.
34
00:06:35.512 --> 00:06:47.432
Uh, we live entirely in the second category. This is a category that's increasingly recognized as central to defense readiness. Uh, the difference isn't necessarily equipment.
35
00:06:47.492 --> 00:06:58.812
Mind you, we have a lot of the equipment replicated in our product. It works. There's medications and a lot of the natural, uh, elements that you would need or anyone would need to run a simulation.
36
00:06:59.492 --> 00:07:08.752
What's really important is how fast and how well humans communicate, think, and act in those simulations. And oftentimes they get some practice, uh,
37
00:07:09.672 --> 00:07:19.932
and they realize through that immersive practice that there's safety, there's psychological safety in the simulation mode. They can try things. It's pretty safe. There's not a lot of downside to it.
38
00:07:20.092 --> 00:07:33.312
Most times, in a self-directed learning mode, an individual can jump in and kinda get warmed up to the experience, and they can try that a few times and then come into the experience with a live trainer and run that with their teammates.
39
00:07:33.392 --> 00:07:59.312
When you think about a medevac crew responding to the multi-casualty incident, uh, example that Raja mentioned here, or some type of combat medic managing a trauma in the field, the difference between a good outcome and a catastrophic one is often, uh, coming down to how well teams communicate with each other, how well they act in decisions that are quite critical in the first ten to fifteen minutes of treatment.
40
00:07:59.892 --> 00:08:15.052
That's how we build that level of competency and execution under pressure. When you think about simulation for Lumeto, we train really around AI-based clinical scenarios that really focus on human dynamics.
41
00:08:15.792 --> 00:08:24.432
Inside the simulation itself, we have a full digital experience, and so there's a lot of data being captured, and you can turn that data into a debrief.
42
00:08:25.872 --> 00:08:35.872
You can respond very quickly in an environment by adjusting the simulation. These are all things that are very realistic and kind of reflect the way people learn in real life.
43
00:08:36.012 --> 00:08:46.182
Uh, moving to a digital experience, we try to mimic as much of that as possible, but give the opportunities to adjust and adapt the learning experience to meet where our learners are at.
44
00:08:47.772 --> 00:08:54.672
Kavi, just as you're talking about this, I have a question for you. Now, keep in mind, I'm novice to this whole experience in technology, but
45
00:08:55.672 --> 00:09:07.352
if you're practicing, it's kind of a practice makes perfect, and simulation is a new environment. I would assume then that there are unique learning outcomes that comes from this experience.
46
00:09:07.582 --> 00:09:19.052
So do you actually then have an opportunity to improve from the simulation training and then add it to what the experience would be like in real life? Yeah.
47
00:09:19.092 --> 00:09:33.532
That's actually quite the incredible part of using an adaptable simulation product like Lumeto's involved. We start off with mimicking and really taking some of the current curriculum and adding it into the platform.
48
00:09:34.032 --> 00:09:39.502
But very quickly, educators realize there's a lot of potential. There's a almost untapped potential.
49
00:09:39.652 --> 00:09:52.932
By doing reps and sets, you can start to realize that not every simulation goes in what we call a golden route or, you know, golden run, where it's exactly as programmed or at least imagined.
50
00:09:53.612 --> 00:10:12.792
You start to make adjustments, like you throw in a heart attack, you throw in a non-responsive patient that is going longer than one would think, even with treatment, and you're also working with a dynamic of multiple learners around, so you can change the roles that one plays in every scenario.
51
00:10:12.892 --> 00:10:14.392
So there's a lot of opportunity
52
00:10:16.232 --> 00:10:37.092
in thinking through a variety of capabilities that are being used for learning, but also just the fact that you have a role-based system here that you can swap and try different things and see it from a different perspective, I think creates a ton of adaptability and just increases the number of scenarios that one can run.
53
00:10:37.132 --> 00:10:39.452
And in fact, the more you run it, the more reps you have.
54
00:10:40.132 --> 00:10:53.602
You're increasing not just your clinical hours in the product, but you're also seeing a lot of low-opportunity cases that may happen, and you've never really had that zero-to-one chance, so now you have that zero-to-one chance of experiencing it first time.
55
00:10:54.692 --> 00:11:01.432
So in a way, your company is benchmarking new learning experiences. You guys are kinda like pioneers.
56
00:11:03.232 --> 00:11:18.592
Well, we see a lot of innovation happening, and most folks that use the product start off, like I said, with a preconceived idea of how to run simulation in the product, and then bas- they generally realize there's a lot more potential once you're, you know, acclimated to VR.
57
00:11:19.152 --> 00:11:30.232
You understand the dynamic nature of the patient AI system. There's a lot of communication capabilities that one typically aren't really exercised in high fidelity with mannequins or even actors.
58
00:11:30.272 --> 00:11:40.952
There's a lot of standardization protocols that one can run. So you start to see just this expanding capability set, and scenarios grow, and they evolve in the simulation platform.
59
00:11:40.972 --> 00:11:53.232
And, you know, it's really interesting to see where educators have taken the product, but they typically start with some of the more simple cases, and then it's incredible to see how much it expands with variability from there.
60
00:11:54.032 --> 00:12:03.672
Yeah, I mean, the growth opportunities and the expansion capabilities for training just sound incredible. Raja, I wanna come over to you now.
61
00:12:03.792 --> 00:12:15.672
There's been a lot of talk about the CAF expanding, both regular force and reserves. What does, um, Lumeto offer an organization that's trying to grow its force quickly while maintaining standards?
62
00:12:17.432 --> 00:12:24.212
Yeah, I think that last word is critical, standards. So growing a force and recruiting is obviously one challenge.
63
00:12:24.852 --> 00:12:39.652
But then when you have these folks, how do you onboard them in a way that is standardized, where you know that one medical field operative has the same skill as the next one, and, uh, in the same situation, will be able to perform at a similar level?
64
00:12:40.272 --> 00:12:51.672
That readiness, that field readiness, combat readiness across units that are geographically distributed, where it's difficult to come to centralized training environments, how can you do that?
65
00:12:52.072 --> 00:13:00.892
Uh, it's very, very difficult today. In fact, we would say that it's because of how difficult it is, it's just you don't get that level of standardization that you would want.
66
00:13:01.392 --> 00:13:05.192
And so this technology really does unlock that.
67
00:13:05.552 --> 00:13:14.412
So you've recruited the troops, you've got them, they're in the reserves, in the reg forces, how do you get them to a standard level of clinical capability but maintain that also?
68
00:13:14.592 --> 00:13:22.511
There's new equipment that gets introduced. There's new protocols that get introduced. There's new environments that we're being deployed to around the world that have different, unique characteristics.
69
00:13:22.952 --> 00:13:31.412
How do you maintain and grow, uh, and a standardized medical field force? And that is where we come in.
70
00:13:31.472 --> 00:13:52.612
So it's really about creating, uh, that scale and efficiency that, uh, Kavi was talking about, like being able to deliver this to anywhere, uh, where somebody is, again, having remote access, having a standardized way to do it, having the data that tells you, okay, this learner or this cohort of learners needs more time in this type of simulation to kinda get up to that standard.
71
00:13:53.092 --> 00:14:04.311
Uh, all of those things become possible when you have, uh, a digitally delivered simulation environment. It's much harder to do in a physical simulation, uh, environment. To us, it's about standardization.
72
00:14:04.372 --> 00:14:22.392
I think that is a absolutely key word, and there's a very clunky, expensive, hard way to do that by flying everyone and doing a physical simulation, and that not only is this more cost efficient, uh, it's also, in certain situations, it produces better outcomes.
73
00:14:22.572 --> 00:14:30.272
Think about trying to simulate a mass casualty event. You'd literally have to recruit what we call standardized patients, like actors, and put a lot of them in a room.
74
00:14:30.372 --> 00:14:32.972
Uh, you know, have them acting out different types of injuries.
75
00:14:33.012 --> 00:14:49.381
Like, we can just create a digital simulation of that with AI-driven characters and disruptions, with real sounds and stress of the environment, uh, and put people in it wherever they are, and give them a chance not just to learn the skills, but to acclimatize to th- those types of stressful environments.
76
00:14:49.792 --> 00:14:56.952
And I think, uh, so that standardization is key, and that doing it in an efficient manner is a bonus. You know, it's incredible.
77
00:14:57.032 --> 00:15:10.292
AI is such a game changer to what you guys are doing, so the standards, the consistency, and it's the very cost-efficient way to create these opportunities for people to train.
78
00:15:11.092 --> 00:15:20.652
Kavi, I wanna just come back to you for a second about the training and what that experience is like. Is there a way that you can walk us through what an actual training experience
79
00:15:22.112 --> 00:15:34.692
looks like, specifically in defense medicine or emergency response context? Sure. So let me pick up on where Raja left off on maybe setting a scene for a high-stress situation.
80
00:15:34.752 --> 00:15:45.372
So let's take a medevac scenario, and imagine you're a combat medic, part of a medevac team. You're in a virtual medevac helicopter,
81
00:15:46.942 --> 00:15:57.392
and it's set out visually and spatially accurate, so you've got a sense of that, uh, realistic, uh, environment, and it's constrained, by the way, because you're, of course, in a helicopter.
82
00:15:57.452 --> 00:16:09.472
You've got your, uh, a real aircraft cabin. Um, you've got a pilot that is managing and communicating with you as well, and you have a couple of teammates joining you remotely in that scenario.
83
00:16:10.952 --> 00:16:22.752
Everything is clinical, clinically accurate, I should say. Uh, medical equipment is there. You've got medications, and so now it's go time. The simulation starts. You have a casualty in front of you
84
00:16:23.812 --> 00:16:34.912
whose vitals are deteriorating, right? You've just received this patient, and you've got really a time-based sort of, uh, situation that you've gotta react with your teammates. It's an AI patient.
85
00:16:35.032 --> 00:16:47.132
This person's capable and very realistic, uh, responding, very capable of responding physiologically, uh, as well, so accurate reactions to pain. Uh, the patient is conscious, responding to your voice.
86
00:16:47.172 --> 00:16:53.452
You're speaking, they're responding, they're hurting. And, uh, you're, of course, having to go through a set of procedures.
87
00:16:53.472 --> 00:17:04.632
You're able to communicate in shorthand to the AI pilot if needed, by the way, and that communication is really important because it's, these are all protocols and new standards in communicating 'cause you're a medevac team.
88
00:17:04.892 --> 00:17:15.132
These are not your traditional comms [clears throat] that you'd be practicing otherwise. Your teammates are reacting to the situation in real time, and you're coordinating treatment.
89
00:17:15.492 --> 00:17:23.932
Every decision that you're making there, whether it's airway management in case of harm to the neck, some type of hemorrhage control,
90
00:17:25.032 --> 00:17:39.192
medication administration, you're applying a tourniquet, everything is being tracked, and that, the way that's tracked is through a system called ACE in Involve XR. ACE stands for Artificial Clinical Evaluator.
91
00:17:39.952 --> 00:17:52.712
Because it's an all-digital environment, we're able to see, hear, we're able to monitor and really capture all data, and this is running a competency assessment in the background through the scenario.
92
00:17:53.072 --> 00:17:58.912
It's individualized. You're seeing, you're capturing data from every individual so you know who's doing what.
93
00:17:58.972 --> 00:18:11.032
You're listening to the conversation, and you're basically monitoring and checking off actions and procedures that are being achieved through that process. Uh, it knows, by the way, the NATO standard for that protocol.
94
00:18:11.092 --> 00:18:19.352
It knows what good looks like, and it knows when there's a delay. It's able to interpret conversation. It's able to interpret actions.
95
00:18:19.712 --> 00:18:33.612
Even if you did the right action and you did it out of sequence, it can recognize that. After the scenario, uh, you're done. It typically takes about ten, fifteen minutes, uh, if you go through a multi-learner scenario.
96
00:18:34.252 --> 00:18:39.752
We have another AI product capability called Artificial Clinical Facilitator, and this one's really important.
97
00:18:40.392 --> 00:18:52.602
Uh, it runs a structured debrief, walking you through the specific moments where your decision-making diverged from best practice. It's really a reflection exercise. Where did you hesitate?
98
00:18:52.712 --> 00:19:04.042
Where did you deviate from protocol? Um, was there something... It's asking you questions. It's having you think and respond to how you were feeling as you were going through that exercise,
99
00:19:05.272 --> 00:19:18.132
and that's really important, that reflection exercise. Simulationists around the world, those that run healthcare training, know that this is where most of the learning actually takes place, is in that reflection.
100
00:19:18.872 --> 00:19:29.912
And so we've developed both the opening of the scenario, the running of the scenario, and the closing of the scenario. The entire cycle can be completed in about twenty-five minutes or so,
101
00:19:31.112 --> 00:19:44.032
and learners find an incredible amount of psychological safety in this exercise. They're able to complete the exercise, and of course, with that level of psychological safety, they want to rerun the scenario.
102
00:19:44.792 --> 00:19:53.272
But here's what happens when you rerun the scenario. We don't want you to go back in and do the exact same thing. You've sort of picked up on a level of progression here in your competency.
103
00:19:53.332 --> 00:20:03.392
We're hoping that the next run is gonna be different, and this is where the adaptability of the platform comes in. The instructor would throw in a twist. They would change the scenario.
104
00:20:03.432 --> 00:20:09.692
They may change, uh, some element of the patient reaction. They may not respond to the medication.
105
00:20:09.752 --> 00:20:17.152
They may throw in, uh, a lack of a medication, take one of the medication opportunities out from the scenario and see how you do, right?
106
00:20:17.212 --> 00:20:24.272
So this is what a typical scenario, uh, would look like, or an experience would look like in the platform, and we've been running it for some time.
107
00:20:24.732 --> 00:20:30.912
Of course, we have fifty-plus hospitals and academic institutions that have been exercising this platform.
108
00:20:30.972 --> 00:20:41.252
It's, you know, sort of tested in, I wouldn't say battle tested in the typical way military defense would look at that, but it's been tested as a software platform, but also from a learning perspective.
109
00:20:41.732 --> 00:20:55.932
We got peer-reviewed validations from NYU, Mayo Clinic, Wake Forest, a lot of top brands of healthcare institutions that use our product and have been validating that this adaptive level of learning makes a huge difference.
110
00:20:55.992 --> 00:21:06.452
So the layer of readiness that's been hardest to measure, that's exactly what we're solving here with, with the simulation. That's really interesting.
111
00:21:07.212 --> 00:21:22.012
Raj, Lumedic has been actively engaging with defense primes around ITB. What makes your company a compelling ITB partner, and how do you think about that role as a company?
112
00:21:24.372 --> 00:21:37.092
Yeah, thanks for that question. IT is something we've been learning about, uh, in the last little while, and as we've dug into it, we're like, "Oh, okay, I think we are a bullseye here," uh, for a few reasons.
113
00:21:37.532 --> 00:21:49.652
You've heard in this conversation all about frontier technology and innovations and R&D that we're doing with AI and how to fundamentally change, uh, experiential learning in healthcare.
114
00:21:50.252 --> 00:22:02.892
That is, by its very nature, an R&D effort. We have been, uh, developing new frontier technology, experimenting with it, figuring out what works and what doesn't work. It's all very AI-driven.
115
00:22:02.912 --> 00:22:08.272
It's all about simulation and learning outcomes. Like, this is a area where we are not only
116
00:22:09.172 --> 00:22:19.832
building and releasing products, but we are constantly, uh, researching and developing and trying new things, and we don't do that in a vacuum either. We partner with academic institutions.
117
00:22:19.912 --> 00:22:32.252
There's been four different IRB-certified studies done on our platform. There's two more in progress right now. So we are engaged in the R&D side of this in a way that is, I think, very ITB-friendly.
118
00:22:32.772 --> 00:22:44.472
Uh, we are a research partner and not just a commercial partner. Uh, and I think the key thing here is that you can see the outcome. It's not just a project, a standalone project.
119
00:22:44.492 --> 00:22:55.532
We are trying to fundamentally change the capability of training and education in healthcare across defense. That is a big, bold goal to have.
120
00:22:55.922 --> 00:23:07.772
And so every little bit of R&D and every little, uh, deployment or every pilot we do is gonna get us further towards that goal. So having that, we're not just a project, uh, that checks a box on an ITB.
121
00:23:07.812 --> 00:23:19.791
We're a real research partner that is gonna help progress towards a, a massive new capability and a new goal for, um, defense organizations. So we see that as a great partnership opportunity.
122
00:23:19.912 --> 00:23:30.872
Uh, we also happen to check all the boxes on the very high multiplier areas in the ITB framework around AI training and simulation and Indigenous and remote community development.
123
00:23:31.352 --> 00:23:40.012
Uh, these are all just things, this is the core of what we do, uh, and they happen to be super high multipliers. We're a Canadian startup with, you know, 30 people doing R&D.
124
00:23:40.092 --> 00:23:49.132
Like, I think we do check a lot of the boxes, but even more importantly than that, we're doing work that is, is gonna have a big impact on how the forces train. Oh, I think it will.
125
00:23:49.612 --> 00:24:05.872
Kavi, on the AI side, you've discussed the ACE and the ACE F systems, and like Raja says, these things are pretty innovative, and they do qualify as R&D. In fact, I think they're genuinely novel and unique.
126
00:24:06.532 --> 00:24:21.132
Tell us a little more about what's hard about building AI that works in this context and, uh, where the R&D frontier is. Yeah, sure. I actually should start by saying a lot of this is indeed novel.
127
00:24:21.452 --> 00:24:25.272
We have some patents pending on the platform capability.
128
00:24:25.332 --> 00:24:40.112
It's truly- What we believe is a, not just the next generation of simulation training, but ultimately a, a modality that we need to protect, not just as a company, but as a capability in Canada and export this capability, uh, as best we can.
129
00:24:40.232 --> 00:24:52.132
So there's a lot of opportunity to leverage the product and capabilities that we're developing here on the platform. There are really two hard areas that we focus on, and just to illustrate
130
00:24:53.232 --> 00:25:08.272
what's been accomplished and tested and validated, we focus on multi-agent complexity and contextual language and just all the communication factors, but also the procedural and the elements of running a simulation and the product.
131
00:25:08.281 --> 00:25:22.052
And, and if you kind of think of the makeup of the environment, you may have a, a single patient normally in a hospital environment, but you apply that to defense, and now you're dealing with multiplicity.
132
00:25:22.092 --> 00:25:24.772
You're dealing with multiple patients, mass casualty.
133
00:25:24.792 --> 00:25:41.332
You may be dealing with a lot more medics working in a non-traditional sense, right, around a room f- full of casualties, and they're having to tag each other and go from person to person and individual to individual and kinda deal with the very different complex situations.
134
00:25:41.392 --> 00:25:55.532
So we wanna get to that level of complexity. So we validated, obviously, single patient simulation rigorously. That's been used. It holds up quite well, both in the modes of a live training environment as well as
135
00:25:56.952 --> 00:26:07.792
running reps and sets, uh, through self-directed learning mode, right? That's where a, a learner can pick up the headset and jump into a scenario and run it as many times as they want, and even that is adaptable.
136
00:26:08.352 --> 00:26:21.722
The frontier is where we are focused, and that is moving to multiple casualties with different acuity levels and different complexities. Having multiple response teams.
137
00:26:21.752 --> 00:26:35.952
Today, we can handle four, five, six learners in the platform at the same time, but then creating the scenarios that make a four-person response team truly productive and truly coordinated is where the complexity lies.
138
00:26:36.021 --> 00:26:46.112
And sure, you're talking with each other, you're talking with the patient, and there's a comms level of complexity in the AI system that we of course have to develop and work out, right?
139
00:26:46.152 --> 00:26:57.512
Those things are not necessarily straightforward, to be able to speak to a real person versus an AI system and also have AI that's tracking all of the data that's occurring in the simulation.
140
00:26:57.972 --> 00:27:06.072
So these things create a set of conditions that can be highly complex architecturally. We wanna make sure it's robust.
141
00:27:06.132 --> 00:27:21.552
We wanna make sure that we're running the LLMs and the simulation of our product in a mode that it realizes that the realistic nature of running these scenarios can be tested beforehand, and that makes it more robust when it's put to work with a scenario.
142
00:27:22.192 --> 00:27:33.112
So that's the architecture conditions that we're facing now. We think it's the right place to spend time and to really develop the next wave of capabilities around this. Uh, there's a language side as well.
143
00:27:33.312 --> 00:27:45.602
The conversational components are important. The virtual patients have multi-language support, so we know that this product is a, highly valuable when it's exportable to our NATO partners.
144
00:27:45.632 --> 00:28:00.812
And we believe with the platform in its current capability, which has about eight languages supported in the product today, these are with the patients, communication, the interface, um, we believe that there's a huge potential to take that and expand it even further and make that more robust.
145
00:28:01.092 --> 00:28:09.052
So there's a deep amount of capabilities, both architecturally with multiple patients, multiple learners, as well as communication. And then, of course,
146
00:28:10.592 --> 00:28:23.852
the AI systems that track and monitor and build the competency tracking for each learner is one of the final parts that is important. That's where there's deep specific training expertise built into the platform.
147
00:28:23.912 --> 00:28:28.812
It's tracking what each and every individual is learning through each of the scenarios.
148
00:28:29.482 --> 00:28:37.672
And the more they use it, the more we can build a preparedness, uh, readiness, uh, checklist and overall graph of each of the individual learners.
149
00:28:37.752 --> 00:28:54.092
So it kinda goes down the entire workflow of starting the simulation, running it with multiple complexities of patients and, and learners, and then ultimately building a competency profile for that, uh, for that learner that one can take and develop further.
150
00:28:55.352 --> 00:29:07.412
And of course, AI capability and technologies are just gonna continue to multiply, which will make this even more, uh, effective. And that leads me into my next question, the final question.
151
00:29:08.052 --> 00:29:23.612
Raja, as you think about the broader audience that's listening to this podcast, the military decision-makers, companies that are in the defense ecosystem, government officials, what do you want each of them to take away from this show?
152
00:29:27.352 --> 00:29:33.852
Yeah, great question. There's probably three different messages for the three different rooms.
153
00:29:33.892 --> 00:29:40.332
For, for the military training and modernization leads, you know, we have a platform that's been proven in clinical environments.
154
00:29:40.932 --> 00:29:49.052
It's in many ways, uh, more demanding than training simulators and mannequins and so on, and more stressful and more realistic.
155
00:29:49.092 --> 00:30:01.532
It's ready to be adapted for defense, emergency response at scale, uh, for all the field clinician in the force, and I think that's a key message for the, you know, the training and modernization leads.
156
00:30:02.212 --> 00:30:07.852
We have AI-driven competencies assessment. We have a lot of different things that can be applied to different use cases.
157
00:30:07.932 --> 00:30:20.352
Honestly, we haven't even uncovered all the different defense use cases where this technology can apply. That is an area of potential partnership and ITBs and piloting that we're very, very excited.
158
00:30:20.412 --> 00:30:32.612
Like we said, this is a transformational change in education for healthcare, and it's gonna take time. For companies in the ecosystem, these are the BAEs and system integrators. You know, we are a small startup.
159
00:30:32.642 --> 00:30:39.652
We're not a competitor. We're an enabler. Uh, we're a research and product partner that's focused on one specific thing.
160
00:30:39.712 --> 00:30:46.344
How do you create better prepared- troops when, in the cases of medical emergencies and emergency response. Like, that's it.
161
00:30:46.424 --> 00:30:53.104
We're doing one thing, and we're doing it, trying to do it a way that's scalable a- and efficient. So those are all potential partners for us.
162
00:30:53.484 --> 00:31:07.324
I think we're looking at ITB obligations, we're looking at AI, R&D, uh, all of those things, we're here, uh, to support you. For government, we're a Canadian company, uh, with Canadian IP.
163
00:31:07.644 --> 00:31:22.144
Uh, we have provisional patents, four provisional patents filed and more to come. We are, I think, what, uh, this government has been talking about. Like, it's a homegrown, innovative, dual use technology startup.
164
00:31:22.184 --> 00:31:32.864
We have customers all over the world, and we're ready to really, really tackle this. I think the key message to government is help us help, you know, the defense sector.
165
00:31:33.204 --> 00:31:41.704
Give us a pathway to be procured that is more, uh, accessible. You- I'm sure you've done podcasts on just that topic of procurement.
166
00:31:41.744 --> 00:31:51.184
It's a, it's a big, complex machine, and there seems to be a lot of will to change it now, which is very exciting. It hasn't landed on our doorstep yet. There's a lot of complex barriers.
167
00:31:51.324 --> 00:31:57.144
Even when we have champions in DND CAF that want to work with us,
168
00:31:58.384 --> 00:32:12.184
they are having a hard time figuring out how to do that, and I think that's a key message to government is, like, just clear the pathway, uh, for companies like ours to really get involved and to start developing, uh, this great technology.
169
00:32:12.284 --> 00:32:19.584
Yeah. I, I agree, and that's certainly the theme in a lot of the conversation that, that we hear these days about clearing the path.
170
00:32:20.424 --> 00:32:31.884
You have a great product with a lot of potential and a lot of opportunity, and it is great to hear, you know, this is a great Canadian success story. You've been listening to Vanguard Radio. I'm your host, J.
171
00:32:32.044 --> 00:32:46.964
Richard Jones. We've covered a lot today. Uh, we've talked a lot about training and simulation. We've talked about how it can be successfully transferred from the classroom to scalable simulation for readiness.
172
00:32:47.504 --> 00:32:58.784
I wanna thank our very, very special guests for coming on the show, uh, Raj Akana, who is the CEO of Lumeto, and Kavi Maharaj, who is the chief product officer, CPO, with Lumeto.
173
00:32:59.164 --> 00:33:05.124
Gentlemen, it's been so interesting, and thank you very much for coming on. Thank you for having us. Thank you very much. It's been a pleasure.
174
00:33:05.804 --> 00:33:24.124
The views, information, or opinions expressed during the Vanguard Weekly Podcast series, CGE Weekly Podcast series, Vanguard Radio Podcast, and CGE Radio Podcast are solely those of the individuals involved, and do not represent those of Vanguard or any affiliated media products, their employees, or owners.
175
00:33:24.164 --> 00:33:34.344
Vanguard Magazine, Canadian Government Executive Magazine, and its affiliates are not responsible for, and do not verify for accuracy, any of the information contained in this podcast series.
176
00:33:34.854 --> 00:33:45.844
This podcast may not be copied or distributed without express written approval of Promotive Communications Inc., PCI. You may not edit, modify, or redistribute this podcast without consent from PCI.
177
00:33:46.184 --> 00:33:54.184
PCI assumes no liability for activities in connection with this podcast or for use of this podcast with any other website, computer, or playing device.
1
00:00:00.100 --> 00:00:18.040
[upbeat music] Welcome back to another episode of Vanguard Radio. I'm your host, J. Richard Jones. It's a pleasure to be with you today, and we're gonna get into some very interesting topics related to healthcare.
2
00:00:18.500 --> 00:00:27.460
I have two very special guests that are joining me, and they have a very progressive company that is incorporating healthcare and technology.
3
00:00:27.920 --> 00:00:35.560
And just before the show, we were chatting a little bit about this, and parts of the discussion were touching around the role of AI.
4
00:00:35.800 --> 00:00:49.040
There's an evolution going on in terms of training, and it's setting an incredibly fast pace in terms of the benefits of being able to improve healthcare training and accessibility.
5
00:00:49.080 --> 00:01:01.060
The other side of it, though, is with technology, the benefit is you're able to take training and education for medical professionals and move it from the class into training at scale.
6
00:01:01.500 --> 00:01:13.400
This provides opportunity for better care and for remote training right across the board. Our guests today joining us are Mr. Raja Khanna, who is the CEO of Lumeto, and Mr.
7
00:01:13.920 --> 00:01:26.440
Kavi Maharaj, who is the chief product officer or CPO with Lumeto. Now, if you're not familiar with Lumeto, they were just recently at our simulation and training forum put on by Vanguard.
8
00:01:27.160 --> 00:01:42.320
They are a Canadian healthcare training and dual-use simulation company whose platform involve XR, and it delivers clinically proven AI-enabled and customizable immersive training for healthcare and defense.
9
00:01:42.750 --> 00:01:55.340
It's pretty neat stuff. It was amazing to see their demonstrations, and they'll probably be at CANSEC this year, too, if you wanna meet with them. So gentlemen, welcome to the show. It's a pleasure to be here.
10
00:01:55.920 --> 00:02:06.120
Thank you for having us. It's great to have you on. And generally, we're gonna talk a little bit about the training, training the future force, really, and the advantages of AI and scalable simulation for readiness.
11
00:02:06.660 --> 00:02:16.040
Raja, I'm gonna jump right in. Lumeto is known as a healthcare simulation company. How should the defense community be thinking about who you are?
12
00:02:18.000 --> 00:02:27.150
Yeah, we've been around for about six years training healthcare professionals in the civilian sense in hospitals and nursing schools and medical schools.
13
00:02:27.900 --> 00:02:42.020
And really at our core, we're using AI to simulate complex medical situations to give the chance for clinicians to basically practice at scale without the need for mannequins or physically being somewhere.
14
00:02:42.590 --> 00:02:47.210
That same paradigm, it applies directly to the defense sector.
15
00:02:47.820 --> 00:03:07.100
Uh, we've been lucky enough to do a few, uh, pilots with the Canadian Armed Forces, and, uh, we see big opportunity here because when you have a distributed, you know, uh, force that all need to have standardized set of skills and all need a chance to practice and gain confidence in complex scenarios that are hard to learn in a classroom, how do you do that?
16
00:03:07.520 --> 00:03:23.380
Uh, we've come up with a solution that is, uh, really about immersive simulation that uses AI to basically mimic these types of environments and the stress and complexity of those types of environments, giving people a chance to practice, learn, build confidence, and build, uh, uh, medical skill.
17
00:03:23.680 --> 00:03:32.560
So that's who we are. We're pretty excited about the dual use approach for our business, taking our, what we've done in the hospital and academic world and bringing it to defense.
18
00:03:34.360 --> 00:03:41.000
How's it been received so far by the defense community, Raja? Uh, it's been overwhelmingly positive.
19
00:03:41.180 --> 00:03:47.460
Uh, I think our use cases and our thesis for our business, we're, we're working in hospitals and academic institutions around the world now.
20
00:03:48.040 --> 00:04:00.900
It fits even better in defense because here you have people that need to be trained that are almost always spread out across, uh, the world, if not a nation. Uh, and this technology works incredibly effectively remotely.
21
00:04:00.920 --> 00:04:13.040
You no longer have to fly people into training centers. It's easier to deploy at scale. Uh, there's more opportunity for training, uh, when you have this type of digital technology that can be adopted anywhere.
22
00:04:13.080 --> 00:04:16.660
You can have an instructor in one city and, you know, learners in another city.
23
00:04:17.220 --> 00:04:34.700
Uh, that's incredibly powerful, and I think just AI unlocks all kinds of new modalities that even if you did fly them all to the same place and train them on a mannequin, which is how it often happens today, that mannequin usually can't talk, or you have to hire actors as standardized patients.
24
00:04:34.820 --> 00:04:48.840
It has been traditionally very difficult to train complex medical scenarios. Now we can create those like a mass casualty event in a simulated AI-driven virtual environment, uh, and it's, uh, pretty compelling.
25
00:04:48.969 --> 00:04:58.420
So I'd say the response has been overwhelmingly positive. Uh, I think the defense community have already adopted these types of modality for other combat training and so on.
26
00:04:58.770 --> 00:05:06.320
And so in healthcare, I think in particular, this is the next, uh, frontier, and, and people are excited to engage. I'm not surprised.
27
00:05:06.400 --> 00:05:18.010
I mean, the technology that is allowing for this type of training and access to go on is just incredible, and remote access, I mean, just imagine all of the different things that can be done.
28
00:05:18.660 --> 00:05:30.620
Kavi, I wanna come over to you now. The term simulation means very different things to people in the defense world. How do you position what Lumeto does with that landscape?
29
00:05:32.240 --> 00:05:34.840
Yeah, that's a great question, and it's an important distinction.
30
00:05:34.920 --> 00:05:51.060
I'll tell you, we first of all fit into synthetic simulation, so we of course know that the military and most training, uh, modalities for simulation today are what's called high fidelity, so that's realistic training with mannequins or sometimes real people, non-scalable.
31
00:05:51.720 --> 00:05:57.860
And within synthetic simulation, there are really two categories that we see happening, uh, and existing today.
32
00:05:58.040 --> 00:06:10.562
One of them is kinetic-based simulation, a lot of familiarity with that, mostly weapon systems and vehicle operation, tactical type, uh, maneuvers. And training programs around that.
33
00:06:10.852 --> 00:06:34.892
We are actually focused on human factor training, and that difference there is mainly, uh, focused around human performance, communication, team coordination, clinical decision-making under stress, and overall just trying to get a sense of confidence in actually running the simulation in a firsthand, first-person experience.
34
00:06:35.512 --> 00:06:47.432
Uh, we live entirely in the second category. This is a category that's increasingly recognized as central to defense readiness. Uh, the difference isn't necessarily equipment.
35
00:06:47.492 --> 00:06:58.812
Mind you, we have a lot of the equipment replicated in our product. It works. There's medications and a lot of the natural, uh, elements that you would need or anyone would need to run a simulation.
36
00:06:59.492 --> 00:07:08.752
What's really important is how fast and how well humans communicate, think, and act in those simulations. And oftentimes they get some practice, uh,
37
00:07:09.672 --> 00:07:19.932
and they realize through that immersive practice that there's safety, there's psychological safety in the simulation mode. They can try things. It's pretty safe. There's not a lot of downside to it.
38
00:07:20.092 --> 00:07:33.312
Most times, in a self-directed learning mode, an individual can jump in and kinda get warmed up to the experience, and they can try that a few times and then come into the experience with a live trainer and run that with their teammates.
39
00:07:33.392 --> 00:07:59.312
When you think about a medevac crew responding to the multi-casualty incident, uh, example that Raja mentioned here, or some type of combat medic managing a trauma in the field, the difference between a good outcome and a catastrophic one is often, uh, coming down to how well teams communicate with each other, how well they act in decisions that are quite critical in the first ten to fifteen minutes of treatment.
40
00:07:59.892 --> 00:08:15.052
That's how we build that level of competency and execution under pressure. When you think about simulation for Lumeto, we train really around AI-based clinical scenarios that really focus on human dynamics.
41
00:08:15.792 --> 00:08:24.432
Inside the simulation itself, we have a full digital experience, and so there's a lot of data being captured, and you can turn that data into a debrief.
42
00:08:25.872 --> 00:08:35.872
You can respond very quickly in an environment by adjusting the simulation. These are all things that are very realistic and kind of reflect the way people learn in real life.
43
00:08:36.012 --> 00:08:46.182
Uh, moving to a digital experience, we try to mimic as much of that as possible, but give the opportunities to adjust and adapt the learning experience to meet where our learners are at.
44
00:08:47.772 --> 00:08:54.672
Kavi, just as you're talking about this, I have a question for you. Now, keep in mind, I'm novice to this whole experience in technology, but
45
00:08:55.672 --> 00:09:07.352
if you're practicing, it's kind of a practice makes perfect, and simulation is a new environment. I would assume then that there are unique learning outcomes that comes from this experience.
46
00:09:07.582 --> 00:09:19.052
So do you actually then have an opportunity to improve from the simulation training and then add it to what the experience would be like in real life? Yeah.
47
00:09:19.092 --> 00:09:33.532
That's actually quite the incredible part of using an adaptable simulation product like Lumeto's involved. We start off with mimicking and really taking some of the current curriculum and adding it into the platform.
48
00:09:34.032 --> 00:09:39.502
But very quickly, educators realize there's a lot of potential. There's a almost untapped potential.
49
00:09:39.652 --> 00:09:52.932
By doing reps and sets, you can start to realize that not every simulation goes in what we call a golden route or, you know, golden run, where it's exactly as programmed or at least imagined.
50
00:09:53.612 --> 00:10:12.792
You start to make adjustments, like you throw in a heart attack, you throw in a non-responsive patient that is going longer than one would think, even with treatment, and you're also working with a dynamic of multiple learners around, so you can change the roles that one plays in every scenario.
51
00:10:12.892 --> 00:10:14.392
So there's a lot of opportunity
52
00:10:16.232 --> 00:10:37.092
in thinking through a variety of capabilities that are being used for learning, but also just the fact that you have a role-based system here that you can swap and try different things and see it from a different perspective, I think creates a ton of adaptability and just increases the number of scenarios that one can run.
53
00:10:37.132 --> 00:10:39.452
And in fact, the more you run it, the more reps you have.
54
00:10:40.132 --> 00:10:53.602
You're increasing not just your clinical hours in the product, but you're also seeing a lot of low-opportunity cases that may happen, and you've never really had that zero-to-one chance, so now you have that zero-to-one chance of experiencing it first time.
55
00:10:54.692 --> 00:11:01.432
So in a way, your company is benchmarking new learning experiences. You guys are kinda like pioneers.
56
00:11:03.232 --> 00:11:18.592
Well, we see a lot of innovation happening, and most folks that use the product start off, like I said, with a preconceived idea of how to run simulation in the product, and then bas- they generally realize there's a lot more potential once you're, you know, acclimated to VR.
57
00:11:19.152 --> 00:11:30.232
You understand the dynamic nature of the patient AI system. There's a lot of communication capabilities that one typically aren't really exercised in high fidelity with mannequins or even actors.
58
00:11:30.272 --> 00:11:40.952
There's a lot of standardization protocols that one can run. So you start to see just this expanding capability set, and scenarios grow, and they evolve in the simulation platform.
59
00:11:40.972 --> 00:11:53.232
And, you know, it's really interesting to see where educators have taken the product, but they typically start with some of the more simple cases, and then it's incredible to see how much it expands with variability from there.
60
00:11:54.032 --> 00:12:03.672
Yeah, I mean, the growth opportunities and the expansion capabilities for training just sound incredible. Raja, I wanna come over to you now.
61
00:12:03.792 --> 00:12:15.672
There's been a lot of talk about the CAF expanding, both regular force and reserves. What does, um, Lumeto offer an organization that's trying to grow its force quickly while maintaining standards?
62
00:12:17.432 --> 00:12:24.212
Yeah, I think that last word is critical, standards. So growing a force and recruiting is obviously one challenge.
63
00:12:24.852 --> 00:12:39.652
But then when you have these folks, how do you onboard them in a way that is standardized, where you know that one medical field operative has the same skill as the next one, and, uh, in the same situation, will be able to perform at a similar level?
64
00:12:40.272 --> 00:12:51.672
That readiness, that field readiness, combat readiness across units that are geographically distributed, where it's difficult to come to centralized training environments, how can you do that?
65
00:12:52.072 --> 00:13:00.892
Uh, it's very, very difficult today. In fact, we would say that it's because of how difficult it is, it's just you don't get that level of standardization that you would want.
66
00:13:01.392 --> 00:13:05.192
And so this technology really does unlock that.
67
00:13:05.552 --> 00:13:14.412
So you've recruited the troops, you've got them, they're in the reserves, in the reg forces, how do you get them to a standard level of clinical capability but maintain that also?
68
00:13:14.592 --> 00:13:22.511
There's new equipment that gets introduced. There's new protocols that get introduced. There's new environments that we're being deployed to around the world that have different, unique characteristics.
69
00:13:22.952 --> 00:13:31.412
How do you maintain and grow, uh, and a standardized medical field force? And that is where we come in.
70
00:13:31.472 --> 00:13:52.612
So it's really about creating, uh, that scale and efficiency that, uh, Kavi was talking about, like being able to deliver this to anywhere, uh, where somebody is, again, having remote access, having a standardized way to do it, having the data that tells you, okay, this learner or this cohort of learners needs more time in this type of simulation to kinda get up to that standard.
71
00:13:53.092 --> 00:14:04.311
Uh, all of those things become possible when you have, uh, a digitally delivered simulation environment. It's much harder to do in a physical simulation, uh, environment. To us, it's about standardization.
72
00:14:04.372 --> 00:14:22.392
I think that is a absolutely key word, and there's a very clunky, expensive, hard way to do that by flying everyone and doing a physical simulation, and that not only is this more cost efficient, uh, it's also, in certain situations, it produces better outcomes.
73
00:14:22.572 --> 00:14:30.272
Think about trying to simulate a mass casualty event. You'd literally have to recruit what we call standardized patients, like actors, and put a lot of them in a room.
74
00:14:30.372 --> 00:14:32.972
Uh, you know, have them acting out different types of injuries.
75
00:14:33.012 --> 00:14:49.381
Like, we can just create a digital simulation of that with AI-driven characters and disruptions, with real sounds and stress of the environment, uh, and put people in it wherever they are, and give them a chance not just to learn the skills, but to acclimatize to th- those types of stressful environments.
76
00:14:49.792 --> 00:14:56.952
And I think, uh, so that standardization is key, and that doing it in an efficient manner is a bonus. You know, it's incredible.
77
00:14:57.032 --> 00:15:10.292
AI is such a game changer to what you guys are doing, so the standards, the consistency, and it's the very cost-efficient way to create these opportunities for people to train.
78
00:15:11.092 --> 00:15:20.652
Kavi, I wanna just come back to you for a second about the training and what that experience is like. Is there a way that you can walk us through what an actual training experience
79
00:15:22.112 --> 00:15:34.692
looks like, specifically in defense medicine or emergency response context? Sure. So let me pick up on where Raja left off on maybe setting a scene for a high-stress situation.
80
00:15:34.752 --> 00:15:45.372
So let's take a medevac scenario, and imagine you're a combat medic, part of a medevac team. You're in a virtual medevac helicopter,
81
00:15:46.942 --> 00:15:57.392
and it's set out visually and spatially accurate, so you've got a sense of that, uh, realistic, uh, environment, and it's constrained, by the way, because you're, of course, in a helicopter.
82
00:15:57.452 --> 00:16:09.472
You've got your, uh, a real aircraft cabin. Um, you've got a pilot that is managing and communicating with you as well, and you have a couple of teammates joining you remotely in that scenario.
83
00:16:10.952 --> 00:16:22.752
Everything is clinical, clinically accurate, I should say. Uh, medical equipment is there. You've got medications, and so now it's go time. The simulation starts. You have a casualty in front of you
84
00:16:23.812 --> 00:16:34.912
whose vitals are deteriorating, right? You've just received this patient, and you've got really a time-based sort of, uh, situation that you've gotta react with your teammates. It's an AI patient.
85
00:16:35.032 --> 00:16:47.132
This person's capable and very realistic, uh, responding, very capable of responding physiologically, uh, as well, so accurate reactions to pain. Uh, the patient is conscious, responding to your voice.
86
00:16:47.172 --> 00:16:53.452
You're speaking, they're responding, they're hurting. And, uh, you're, of course, having to go through a set of procedures.
87
00:16:53.472 --> 00:17:04.632
You're able to communicate in shorthand to the AI pilot if needed, by the way, and that communication is really important because it's, these are all protocols and new standards in communicating 'cause you're a medevac team.
88
00:17:04.892 --> 00:17:15.132
These are not your traditional comms [clears throat] that you'd be practicing otherwise. Your teammates are reacting to the situation in real time, and you're coordinating treatment.
89
00:17:15.492 --> 00:17:23.932
Every decision that you're making there, whether it's airway management in case of harm to the neck, some type of hemorrhage control,
90
00:17:25.032 --> 00:17:39.192
medication administration, you're applying a tourniquet, everything is being tracked, and that, the way that's tracked is through a system called ACE in Involve XR. ACE stands for Artificial Clinical Evaluator.
91
00:17:39.952 --> 00:17:52.712
Because it's an all-digital environment, we're able to see, hear, we're able to monitor and really capture all data, and this is running a competency assessment in the background through the scenario.
92
00:17:53.072 --> 00:17:58.912
It's individualized. You're seeing, you're capturing data from every individual so you know who's doing what.
93
00:17:58.972 --> 00:18:11.032
You're listening to the conversation, and you're basically monitoring and checking off actions and procedures that are being achieved through that process. Uh, it knows, by the way, the NATO standard for that protocol.
94
00:18:11.092 --> 00:18:19.352
It knows what good looks like, and it knows when there's a delay. It's able to interpret conversation. It's able to interpret actions.
95
00:18:19.712 --> 00:18:33.612
Even if you did the right action and you did it out of sequence, it can recognize that. After the scenario, uh, you're done. It typically takes about ten, fifteen minutes, uh, if you go through a multi-learner scenario.
96
00:18:34.252 --> 00:18:39.752
We have another AI product capability called Artificial Clinical Facilitator, and this one's really important.
97
00:18:40.392 --> 00:18:52.602
Uh, it runs a structured debrief, walking you through the specific moments where your decision-making diverged from best practice. It's really a reflection exercise. Where did you hesitate?
98
00:18:52.712 --> 00:19:04.042
Where did you deviate from protocol? Um, was there something... It's asking you questions. It's having you think and respond to how you were feeling as you were going through that exercise,
99
00:19:05.272 --> 00:19:18.132
and that's really important, that reflection exercise. Simulationists around the world, those that run healthcare training, know that this is where most of the learning actually takes place, is in that reflection.
100
00:19:18.872 --> 00:19:29.912
And so we've developed both the opening of the scenario, the running of the scenario, and the closing of the scenario. The entire cycle can be completed in about twenty-five minutes or so,
101
00:19:31.112 --> 00:19:44.032
and learners find an incredible amount of psychological safety in this exercise. They're able to complete the exercise, and of course, with that level of psychological safety, they want to rerun the scenario.
102
00:19:44.792 --> 00:19:53.272
But here's what happens when you rerun the scenario. We don't want you to go back in and do the exact same thing. You've sort of picked up on a level of progression here in your competency.
103
00:19:53.332 --> 00:20:03.392
We're hoping that the next run is gonna be different, and this is where the adaptability of the platform comes in. The instructor would throw in a twist. They would change the scenario.
104
00:20:03.432 --> 00:20:09.692
They may change, uh, some element of the patient reaction. They may not respond to the medication.
105
00:20:09.752 --> 00:20:17.152
They may throw in, uh, a lack of a medication, take one of the medication opportunities out from the scenario and see how you do, right?
106
00:20:17.212 --> 00:20:24.272
So this is what a typical scenario, uh, would look like, or an experience would look like in the platform, and we've been running it for some time.
107
00:20:24.732 --> 00:20:30.912
Of course, we have fifty-plus hospitals and academic institutions that have been exercising this platform.
108
00:20:30.972 --> 00:20:41.252
It's, you know, sort of tested in, I wouldn't say battle tested in the typical way military defense would look at that, but it's been tested as a software platform, but also from a learning perspective.
109
00:20:41.732 --> 00:20:55.932
We got peer-reviewed validations from NYU, Mayo Clinic, Wake Forest, a lot of top brands of healthcare institutions that use our product and have been validating that this adaptive level of learning makes a huge difference.
110
00:20:55.992 --> 00:21:06.452
So the layer of readiness that's been hardest to measure, that's exactly what we're solving here with, with the simulation. That's really interesting.
111
00:21:07.212 --> 00:21:22.012
Raj, Lumedic has been actively engaging with defense primes around ITB. What makes your company a compelling ITB partner, and how do you think about that role as a company?
112
00:21:24.372 --> 00:21:37.092
Yeah, thanks for that question. IT is something we've been learning about, uh, in the last little while, and as we've dug into it, we're like, "Oh, okay, I think we are a bullseye here," uh, for a few reasons.
113
00:21:37.532 --> 00:21:49.652
You've heard in this conversation all about frontier technology and innovations and R&D that we're doing with AI and how to fundamentally change, uh, experiential learning in healthcare.
114
00:21:50.252 --> 00:22:02.892
That is, by its very nature, an R&D effort. We have been, uh, developing new frontier technology, experimenting with it, figuring out what works and what doesn't work. It's all very AI-driven.
115
00:22:02.912 --> 00:22:08.272
It's all about simulation and learning outcomes. Like, this is a area where we are not only
116
00:22:09.172 --> 00:22:19.832
building and releasing products, but we are constantly, uh, researching and developing and trying new things, and we don't do that in a vacuum either. We partner with academic institutions.
117
00:22:19.912 --> 00:22:32.252
There's been four different IRB-certified studies done on our platform. There's two more in progress right now. So we are engaged in the R&D side of this in a way that is, I think, very ITB-friendly.
118
00:22:32.772 --> 00:22:44.472
Uh, we are a research partner and not just a commercial partner. Uh, and I think the key thing here is that you can see the outcome. It's not just a project, a standalone project.
119
00:22:44.492 --> 00:22:55.532
We are trying to fundamentally change the capability of training and education in healthcare across defense. That is a big, bold goal to have.
120
00:22:55.922 --> 00:23:07.772
And so every little bit of R&D and every little, uh, deployment or every pilot we do is gonna get us further towards that goal. So having that, we're not just a project, uh, that checks a box on an ITB.
121
00:23:07.812 --> 00:23:19.791
We're a real research partner that is gonna help progress towards a, a massive new capability and a new goal for, um, defense organizations. So we see that as a great partnership opportunity.
122
00:23:19.912 --> 00:23:30.872
Uh, we also happen to check all the boxes on the very high multiplier areas in the ITB framework around AI training and simulation and Indigenous and remote community development.
123
00:23:31.352 --> 00:23:40.012
Uh, these are all just things, this is the core of what we do, uh, and they happen to be super high multipliers. We're a Canadian startup with, you know, 30 people doing R&D.
124
00:23:40.092 --> 00:23:49.132
Like, I think we do check a lot of the boxes, but even more importantly than that, we're doing work that is, is gonna have a big impact on how the forces train. Oh, I think it will.
125
00:23:49.612 --> 00:24:05.872
Kavi, on the AI side, you've discussed the ACE and the ACE F systems, and like Raja says, these things are pretty innovative, and they do qualify as R&D. In fact, I think they're genuinely novel and unique.
126
00:24:06.532 --> 00:24:21.132
Tell us a little more about what's hard about building AI that works in this context and, uh, where the R&D frontier is. Yeah, sure. I actually should start by saying a lot of this is indeed novel.
127
00:24:21.452 --> 00:24:25.272
We have some patents pending on the platform capability.
128
00:24:25.332 --> 00:24:40.112
It's truly- What we believe is a, not just the next generation of simulation training, but ultimately a, a modality that we need to protect, not just as a company, but as a capability in Canada and export this capability, uh, as best we can.
129
00:24:40.232 --> 00:24:52.132
So there's a lot of opportunity to leverage the product and capabilities that we're developing here on the platform. There are really two hard areas that we focus on, and just to illustrate
130
00:24:53.232 --> 00:25:08.272
what's been accomplished and tested and validated, we focus on multi-agent complexity and contextual language and just all the communication factors, but also the procedural and the elements of running a simulation and the product.
131
00:25:08.281 --> 00:25:22.052
And, and if you kind of think of the makeup of the environment, you may have a, a single patient normally in a hospital environment, but you apply that to defense, and now you're dealing with multiplicity.
132
00:25:22.092 --> 00:25:24.772
You're dealing with multiple patients, mass casualty.
133
00:25:24.792 --> 00:25:41.332
You may be dealing with a lot more medics working in a non-traditional sense, right, around a room f- full of casualties, and they're having to tag each other and go from person to person and individual to individual and kinda deal with the very different complex situations.
134
00:25:41.392 --> 00:25:55.532
So we wanna get to that level of complexity. So we validated, obviously, single patient simulation rigorously. That's been used. It holds up quite well, both in the modes of a live training environment as well as
135
00:25:56.952 --> 00:26:07.792
running reps and sets, uh, through self-directed learning mode, right? That's where a, a learner can pick up the headset and jump into a scenario and run it as many times as they want, and even that is adaptable.
136
00:26:08.352 --> 00:26:21.722
The frontier is where we are focused, and that is moving to multiple casualties with different acuity levels and different complexities. Having multiple response teams.
137
00:26:21.752 --> 00:26:35.952
Today, we can handle four, five, six learners in the platform at the same time, but then creating the scenarios that make a four-person response team truly productive and truly coordinated is where the complexity lies.
138
00:26:36.021 --> 00:26:46.112
And sure, you're talking with each other, you're talking with the patient, and there's a comms level of complexity in the AI system that we of course have to develop and work out, right?
139
00:26:46.152 --> 00:26:57.512
Those things are not necessarily straightforward, to be able to speak to a real person versus an AI system and also have AI that's tracking all of the data that's occurring in the simulation.
140
00:26:57.972 --> 00:27:06.072
So these things create a set of conditions that can be highly complex architecturally. We wanna make sure it's robust.
141
00:27:06.132 --> 00:27:21.552
We wanna make sure that we're running the LLMs and the simulation of our product in a mode that it realizes that the realistic nature of running these scenarios can be tested beforehand, and that makes it more robust when it's put to work with a scenario.
142
00:27:22.192 --> 00:27:33.112
So that's the architecture conditions that we're facing now. We think it's the right place to spend time and to really develop the next wave of capabilities around this. Uh, there's a language side as well.
143
00:27:33.312 --> 00:27:45.602
The conversational components are important. The virtual patients have multi-language support, so we know that this product is a, highly valuable when it's exportable to our NATO partners.
144
00:27:45.632 --> 00:28:00.812
And we believe with the platform in its current capability, which has about eight languages supported in the product today, these are with the patients, communication, the interface, um, we believe that there's a huge potential to take that and expand it even further and make that more robust.
145
00:28:01.092 --> 00:28:09.052
So there's a deep amount of capabilities, both architecturally with multiple patients, multiple learners, as well as communication. And then, of course,
146
00:28:10.592 --> 00:28:23.852
the AI systems that track and monitor and build the competency tracking for each learner is one of the final parts that is important. That's where there's deep specific training expertise built into the platform.
147
00:28:23.912 --> 00:28:28.812
It's tracking what each and every individual is learning through each of the scenarios.
148
00:28:29.482 --> 00:28:37.672
And the more they use it, the more we can build a preparedness, uh, readiness, uh, checklist and overall graph of each of the individual learners.
149
00:28:37.752 --> 00:28:54.092
So it kinda goes down the entire workflow of starting the simulation, running it with multiple complexities of patients and, and learners, and then ultimately building a competency profile for that, uh, for that learner that one can take and develop further.
150
00:28:55.352 --> 00:29:07.412
And of course, AI capability and technologies are just gonna continue to multiply, which will make this even more, uh, effective. And that leads me into my next question, the final question.
151
00:29:08.052 --> 00:29:23.612
Raja, as you think about the broader audience that's listening to this podcast, the military decision-makers, companies that are in the defense ecosystem, government officials, what do you want each of them to take away from this show?
152
00:29:27.352 --> 00:29:33.852
Yeah, great question. There's probably three different messages for the three different rooms.
153
00:29:33.892 --> 00:29:40.332
For, for the military training and modernization leads, you know, we have a platform that's been proven in clinical environments.
154
00:29:40.932 --> 00:29:49.052
It's in many ways, uh, more demanding than training simulators and mannequins and so on, and more stressful and more realistic.
155
00:29:49.092 --> 00:30:01.532
It's ready to be adapted for defense, emergency response at scale, uh, for all the field clinician in the force, and I think that's a key message for the, you know, the training and modernization leads.
156
00:30:02.212 --> 00:30:07.852
We have AI-driven competencies assessment. We have a lot of different things that can be applied to different use cases.
157
00:30:07.932 --> 00:30:20.352
Honestly, we haven't even uncovered all the different defense use cases where this technology can apply. That is an area of potential partnership and ITBs and piloting that we're very, very excited.
158
00:30:20.412 --> 00:30:32.612
Like we said, this is a transformational change in education for healthcare, and it's gonna take time. For companies in the ecosystem, these are the BAEs and system integrators. You know, we are a small startup.
159
00:30:32.642 --> 00:30:39.652
We're not a competitor. We're an enabler. Uh, we're a research and product partner that's focused on one specific thing.
160
00:30:39.712 --> 00:30:46.344
How do you create better prepared- troops when, in the cases of medical emergencies and emergency response. Like, that's it.
161
00:30:46.424 --> 00:30:53.104
We're doing one thing, and we're doing it, trying to do it a way that's scalable a- and efficient. So those are all potential partners for us.
162
00:30:53.484 --> 00:31:07.324
I think we're looking at ITB obligations, we're looking at AI, R&D, uh, all of those things, we're here, uh, to support you. For government, we're a Canadian company, uh, with Canadian IP.
163
00:31:07.644 --> 00:31:22.144
Uh, we have provisional patents, four provisional patents filed and more to come. We are, I think, what, uh, this government has been talking about. Like, it's a homegrown, innovative, dual use technology startup.
164
00:31:22.184 --> 00:31:32.864
We have customers all over the world, and we're ready to really, really tackle this. I think the key message to government is help us help, you know, the defense sector.
165
00:31:33.204 --> 00:31:41.704
Give us a pathway to be procured that is more, uh, accessible. You- I'm sure you've done podcasts on just that topic of procurement.
166
00:31:41.744 --> 00:31:51.184
It's a, it's a big, complex machine, and there seems to be a lot of will to change it now, which is very exciting. It hasn't landed on our doorstep yet. There's a lot of complex barriers.
167
00:31:51.324 --> 00:31:57.144
Even when we have champions in DND CAF that want to work with us,
168
00:31:58.384 --> 00:32:12.184
they are having a hard time figuring out how to do that, and I think that's a key message to government is, like, just clear the pathway, uh, for companies like ours to really get involved and to start developing, uh, this great technology.
169
00:32:12.284 --> 00:32:19.584
Yeah. I, I agree, and that's certainly the theme in a lot of the conversation that, that we hear these days about clearing the path.
170
00:32:20.424 --> 00:32:31.884
You have a great product with a lot of potential and a lot of opportunity, and it is great to hear, you know, this is a great Canadian success story. You've been listening to Vanguard Radio. I'm your host, J.
171
00:32:32.044 --> 00:32:46.964
Richard Jones. We've covered a lot today. Uh, we've talked a lot about training and simulation. We've talked about how it can be successfully transferred from the classroom to scalable simulation for readiness.
172
00:32:47.504 --> 00:32:58.784
I wanna thank our very, very special guests for coming on the show, uh, Raj Akana, who is the CEO of Lumeto, and Kavi Maharaj, who is the chief product officer, CPO, with Lumeto.
173
00:32:59.164 --> 00:33:05.124
Gentlemen, it's been so interesting, and thank you very much for coming on. Thank you for having us. Thank you very much. It's been a pleasure.
174
00:33:05.804 --> 00:33:24.124
The views, information, or opinions expressed during the Vanguard Weekly Podcast series, CGE Weekly Podcast series, Vanguard Radio Podcast, and CGE Radio Podcast are solely those of the individuals involved, and do not represent those of Vanguard or any affiliated media products, their employees, or owners.
175
00:33:24.164 --> 00:33:34.344
Vanguard Magazine, Canadian Government Executive Magazine, and its affiliates are not responsible for, and do not verify for accuracy, any of the information contained in this podcast series.
176
00:33:34.854 --> 00:33:45.844
This podcast may not be copied or distributed without express written approval of Promotive Communications Inc., PCI. You may not edit, modify, or redistribute this podcast without consent from PCI.
177
00:33:46.184 --> 00:33:54.184
PCI assumes no liability for activities in connection with this podcast or for use of this podcast with any other website, computer, or playing device.
