From Physics-Based Simulation to Surgical Robots| Show Me The Evidence E8- Dr Dwight Meglan

1
00:00:07,120 --> 00:00:11,800
In the late 1990s, Dr. Dwight
Meglan and I met at

2
00:00:12,000 --> 00:00:13,960
Medicine Meets Virtual
Reality

3
00:00:14,000 --> 00:00:16,000
somewhere in California,
I seem to recall.

4
00:00:17,079 --> 00:00:21,079
Dwight's an engineer. Initially,
when I met him, he was

5
00:00:21,079 --> 00:00:23,800
working on physics-based virtual
reality simulation.

6
00:00:24,239 --> 00:00:28,039
He subsequently moved into
surgical robotics.

7
00:00:28,879 --> 00:00:31,039
You're very welcome
this evening,

8
00:00:31,359 --> 00:00:33,159
Dwight, and nice to
talk to you.

9
00:00:34,159 --> 00:00:37,280
So you developed one
of the first

10
00:00:37,280 --> 00:00:39,560
physics-based virtual reality
simulators.

11
00:00:39,640 --> 00:00:43,880
I purchased one in 1999 at
some considerable cost.

12
00:00:44,560 --> 00:00:49,119
So at that time, how did you
envision that the VR

13
00:00:49,119 --> 00:00:51,439
simulation market would
develop?

14
00:00:52,039 --> 00:00:54,960
And has it developed the way
that you thought it would?

15
00:00:57,200 --> 00:00:59,880
I have to be honest, I really didn't
think about it a whole

16
00:00:59,880 --> 00:01:01,439
lot about relative
to a market.

17
00:01:01,479 --> 00:01:03,759
I was more thinking about it
as a technical problem.

18
00:01:05,159 --> 00:01:08,000
I was just interested in the
idea of how do you do

19
00:01:08,000 --> 00:01:10,239
real-time physics of
tool tissue

20
00:01:10,239 --> 00:01:14,719
interaction and replicate the
real-time fluoroscopy

21
00:01:15,040 --> 00:01:17,200
and give haptic feedback
in a way

22
00:01:17,200 --> 00:01:20,319
that was natural to the
procedure itself.

23
00:01:20,799 --> 00:01:25,079
And I was much more focused on
that than anything else.

24
00:01:25,079 --> 00:01:27,359
And I really didn't give it a
lot of thought other than

25
00:01:27,920 --> 00:01:29,400
there's a lot of catheterization.

26
00:01:29,439 --> 00:01:30,480
So I figured it was
probably going

27
00:01:30,480 --> 00:01:33,159
to be useful and people
would use it.

28
00:01:34,280 --> 00:01:37,439
And do you think that the current
developers of virtual

29
00:01:37,439 --> 00:01:41,959
reality simulation pays much attention
to the physics-based

30
00:01:42,680 --> 00:01:44,120
virtual reality simulation

31
00:01:44,120 --> 00:01:47,040
today, as you did back
in the 1990s?

32
00:01:48,799 --> 00:01:51,480
I don't really think many of
the simulators pay much

33
00:01:51,480 --> 00:01:53,640
attention to that, no matter
what it is.

34
00:01:54,439 --> 00:01:59,599
I was kind of obsessed with
trying to mimic the actual

35
00:01:59,599 --> 00:02:01,560
physics of the tool tissue
interaction.

36
00:02:02,079 --> 00:02:03,920
And I figured if I could get
that right, then the

37
00:02:03,920 --> 00:02:05,599
experience for the user would

38
00:02:05,719 --> 00:02:07,719
reflect reality to
some degree.

39
00:02:08,159 --> 00:02:12,680
So I kind of focused on that.
Whereas, as you well know, a

40
00:02:12,680 --> 00:02:14,000
lot of the simulators kind of

41
00:02:14,000 --> 00:02:16,960
focus on the user experience
itself.

42
00:02:17,479 --> 00:02:19,800
I actually didn't really think
about the user experience.

43
00:02:19,800 --> 00:02:21,800
I just figured if I did
the simulation

44
00:02:21,800 --> 00:02:23,759
right, then the user experience
would be right.

45
00:02:24,840 --> 00:02:28,680
So what do you think have been
the drivers of the current

46
00:02:28,680 --> 00:02:31,199
state of virtual reality
simulation today?

47
00:02:31,919 --> 00:02:35,039
Well, over the last 20 years,
20-odd years?

48
00:02:36,599 --> 00:02:39,680
Basically, the device manufacturers.
I mean, I figured

49
00:02:39,919 --> 00:02:41,120
early on, I didn't
really give

50
00:02:41,120 --> 00:02:42,759
that a lot of thought,
to be honest.

51
00:02:42,759 --> 00:02:46,560
I was just thinking about it
relative to build something

52
00:02:46,560 --> 00:02:48,240
that's useful for the
clinicians.

53
00:02:48,240 --> 00:02:51,159
And that would be something
that would be meaningful.

54
00:02:52,080 --> 00:02:55,439
But as I spent more time working
in the area, because I

55
00:02:55,439 --> 00:02:57,840
worked on other simulators
besides endovascular,

56
00:02:58,280 --> 00:03:01,120
I came to understand that really
the only people that are

57
00:03:01,120 --> 00:03:03,960
really paying for this and really
caring about it a whole

58
00:03:03,960 --> 00:03:06,599
lot were the device
manufacturers.

59
00:03:07,080 --> 00:03:12,960
And they kind of drive the way
things work. And that's kind

60
00:03:12,960 --> 00:03:14,759
of what it seems like it
is today to me.

61
00:03:14,800 --> 00:03:17,719
I mean, there's various professional
societies that have

62
00:03:18,240 --> 00:03:22,199
set up standards for learning
and other things.

63
00:03:23,000 --> 00:03:26,479
But as far as I know, like the
thing I was driving towards

64
00:03:26,520 --> 00:03:29,159
years ago, like 20,
25 years ago,

65
00:03:29,599 --> 00:03:32,639
was trying to figure out how to
come up with physics tests

66
00:03:33,000 --> 00:03:35,759
that would allow you
to verify

67
00:03:35,800 --> 00:03:37,319
the correctness of the
simulator.

68
00:03:38,280 --> 00:03:41,639
And those still don't exist for
the most part, as far as I

69
00:03:41,639 --> 00:03:43,479
know, which I'm kind of
surprised at.

70
00:03:44,479 --> 00:03:47,120
Yeah, but one of the things that
always struck me about the

71
00:03:47,520 --> 00:03:50,319
sensation that the or
the perceptions

72
00:03:50,319 --> 00:03:55,000
are that the user actually had
from the simulation. People

73
00:03:55,000 --> 00:03:57,120
have different sensory
thresholds.

74
00:03:57,479 --> 00:03:59,479
Therefore, the sensory
thresholds are

75
00:03:59,479 --> 00:04:01,400
going to impact on what
they perceive.

76
00:04:01,800 --> 00:04:03,719
And one of the one of the interesting
things I find out

77
00:04:03,719 --> 00:04:07,280
over the years was the plastic
surgeons or microsurgeons.

78
00:04:07,719 --> 00:04:10,319
They won't go for a run. They
won't have coffee.

79
00:04:10,599 --> 00:04:12,120
They try to get a good
night's sleep the

80
00:04:12,120 --> 00:04:14,159
night before they operate
because they know

81
00:04:14,599 --> 00:04:17,639
that they can, a slight
nudge can

82
00:04:17,720 --> 00:04:20,199
impact their performance
the following day.

83
00:04:20,439 --> 00:04:23,279
So it's very hard to actually
come up with, you know, what

84
00:04:23,279 --> 00:04:25,839
are the sensory thresholds for
the clinicians to feel?

85
00:04:26,240 --> 00:04:29,319
So how did you approach that
or try and solve it?

86
00:04:29,879 --> 00:04:32,600
Well, so we did do a couple of
things early on because we

87
00:04:32,600 --> 00:04:35,199
were kind of curious about
how would we

88
00:04:35,759 --> 00:04:37,680
know if we were producing

89
00:04:37,680 --> 00:04:41,920
things that the cardiologists
felt.

90
00:04:43,319 --> 00:04:46,680
And what we ended up doing was
we built an instrumented

91
00:04:46,680 --> 00:04:49,639
torquer that the cardiologists
would

92
00:04:49,639 --> 00:04:51,279
hold to actually manipulate
the tools.

93
00:04:51,879 --> 00:04:54,959
And we figured if we could
measure what

94
00:04:54,959 --> 00:04:57,000
they were actually feeling
in real life,

95
00:04:58,079 --> 00:05:00,560
then we could actually, and we
knew what we were feeding

96
00:05:00,560 --> 00:05:01,920
back to them on the
simulator,

97
00:05:01,920 --> 00:05:04,839
that if they were at least somewhat
correlated, that we

98
00:05:04,839 --> 00:05:07,399
were actually doing at least
some level of a good job.

99
00:05:07,920 --> 00:05:10,879
We never went far enough along
with that to actually try

100
00:05:10,879 --> 00:05:12,000
and figure out if there's

101
00:05:12,000 --> 00:05:13,600
different thresholds for
different people,

102
00:05:13,600 --> 00:05:15,839
which of course there is. We
know from haptics that

103
00:05:16,720 --> 00:05:17,600
different people have just

104
00:05:17,600 --> 00:05:19,319
noticeable difference
thresholds.

105
00:05:20,160 --> 00:05:22,920
But we never pursued
that. I mean,

106
00:05:23,480 --> 00:05:25,639
this was never done
academically.

107
00:05:25,720 --> 00:05:28,560
This was all done in the
auspices of a company.

108
00:05:29,399 --> 00:05:35,319
And since I'm not super active
in the field anymore, like

109
00:05:35,319 --> 00:05:37,680
you said earlier, I mostly work
on surgical robots now.

110
00:05:38,360 --> 00:05:42,000
I'm not aware of anybody having
published extensive data

111
00:05:42,000 --> 00:05:43,959
sets on what it is that a

112
00:05:43,959 --> 00:05:46,360
cardiologist feels when they
do catheterisation.

113
00:05:46,879 --> 00:05:49,399
So there's not minimal
data out there

114
00:05:49,399 --> 00:05:51,360
to even test against,
quite frankly.

115
00:05:51,839 --> 00:05:54,920
And we certainly don't know
much about the actual

116
00:05:54,959 --> 00:05:57,800
interaction of the catheters
with the tissue,

117
00:05:58,240 --> 00:06:00,839
which is what I was really headed
toward. I really wanted

118
00:06:00,839 --> 00:06:02,639
to eventually be able
to say that

119
00:06:03,360 --> 00:06:07,560
we're actually generating the
same forces, contact forces

120
00:06:07,560 --> 00:06:09,199
and pressures and
motions that

121
00:06:09,199 --> 00:06:11,240
you see in real life
quantitatively.

122
00:06:11,759 --> 00:06:15,000
Yeah, I think that's still
a real problem today.

123
00:06:15,879 --> 00:06:20,199
Sometimes I had four of the simulators
that you built when

124
00:06:20,199 --> 00:06:22,399
I worked in University
College Cork.

125
00:06:22,879 --> 00:06:27,240
And occasionally the haptics
weren't switched on and the

126
00:06:27,240 --> 00:06:28,600
clinicians would be
telling me

127
00:06:28,680 --> 00:06:30,360
only the haptics in this
are great.

128
00:06:30,800 --> 00:06:33,560
And I'm not sure what's
going on

129
00:06:33,560 --> 00:06:36,800
here. I mean, I know
about that.

130
00:06:38,600 --> 00:06:43,199
I think over the years, I think
it's just a personal

131
00:06:43,240 --> 00:06:44,399
conclusion that I've come to.

132
00:06:44,680 --> 00:06:46,519
Is it the better the
clinician is, the

133
00:06:46,519 --> 00:06:48,800
more attentional capacity
that they have,

134
00:06:49,199 --> 00:06:52,079
the greater awareness
they have to the

135
00:06:52,079 --> 00:06:56,040
subtle aspects of catheter
wire performance?

136
00:06:56,360 --> 00:06:59,040
And I think it's the same in

137
00:06:59,040 --> 00:07:01,199
robotics or actually
there's no.

138
00:07:06,720 --> 00:07:10,319
Well, except for in the DV5
now does have it.

139
00:07:10,319 --> 00:07:13,319
So, and I've never used it, so
I don't know what it's like.

140
00:07:13,800 --> 00:07:14,000
So.

141
00:07:16,120 --> 00:07:19,000
So you think one of the
biggest drivers

142
00:07:19,040 --> 00:07:22,480
of the physics based virtual
reality simulation

143
00:07:22,480 --> 00:07:24,360
was primarily the
manufacturers.

144
00:07:24,959 --> 00:07:27,680
I mean, and the conclusion
or the observation

145
00:07:27,680 --> 00:07:30,600
I've sort of made is they
seem to want it only

146
00:07:30,879 --> 00:07:35,319
for mostly for marketing and
that education and training

147
00:07:35,360 --> 00:07:37,560
seem to come well down the
track after that.

148
00:07:38,079 --> 00:07:38,959
Would you agree with that?

149
00:07:40,800 --> 00:07:44,480
I mean, everything's subservient
to a purpose

150
00:07:44,480 --> 00:07:46,480
when it's a product, right?

151
00:07:46,480 --> 00:07:48,120
I mean, they're trying
to sell stuff

152
00:07:48,120 --> 00:07:49,439
and that's just the
way it is.

153
00:07:50,000 --> 00:07:51,480
They'd like you to
use it well,

154
00:07:51,879 --> 00:07:53,199
or at the very least
they'd like you

155
00:07:53,199 --> 00:07:54,759
to not use it dangerously,

156
00:07:54,879 --> 00:07:57,600
because if nothing else, it's
bad on their reputation

157
00:07:57,600 --> 00:07:59,199
and maybe it can be more
than that,

158
00:07:59,199 --> 00:08:00,439
depending on the situation.

159
00:08:00,959 --> 00:08:03,519
Could get them into liability
issues or whatever.

160
00:08:03,560 --> 00:08:05,519
So they want you to use
it in a way that

161
00:08:06,000 --> 00:08:09,879
they consider you to be competent
at running the device.

162
00:08:11,199 --> 00:08:16,519
But I mean, I honestly, having
worked on the simulators

163
00:08:16,519 --> 00:08:20,399
and on the robots, I mean,
there's a threshold

164
00:08:20,399 --> 00:08:22,240
at which they kind of
draw the line.

165
00:08:23,199 --> 00:08:26,199
They just want you to be able
to use the tool

166
00:08:26,199 --> 00:08:28,439
in the way that they had
designed it to be used,

167
00:08:28,439 --> 00:08:29,759
at least to their best
understanding

168
00:08:29,759 --> 00:08:31,319
of how it designed it
to be used,

169
00:08:32,200 --> 00:08:35,159
but not, they don't,
as you know,

170
00:08:35,159 --> 00:08:37,600
they don't claim any
responsibility

171
00:08:37,600 --> 00:08:38,919
for the practice of medicine.

172
00:08:38,919 --> 00:08:41,360
So they're not trying to teach
you how to do the procedure.

173
00:08:41,759 --> 00:08:43,879
They're just trying to teach
you how to use the device.

174
00:08:44,000 --> 00:08:45,960
So that's the cutoff.

175
00:08:46,840 --> 00:08:49,679
I'm not entirely sure, but
I had a conversation

176
00:08:49,679 --> 00:08:51,879
with a manufacturer
last week,

177
00:08:52,440 --> 00:08:56,320
and it was about the use of
a stapling device.

178
00:08:56,960 --> 00:08:57,159
Okay.

179
00:08:57,200 --> 00:08:59,639
And the stapling device
has a six to 27%

180
00:09:00,039 --> 00:09:01,320
leak and complication rate.

181
00:09:02,080 --> 00:09:05,799
And one third of the patients that
develop a leak will die.

182
00:09:07,320 --> 00:09:10,080
That's huge consequences
for a relatively--

183
00:09:10,759 --> 00:09:12,960
And the conversation that we're
having now saying that

184
00:09:13,240 --> 00:09:17,200
basically what you want to
be able to do is,

185
00:09:17,279 --> 00:09:19,480
now, well, number one, we
have good evidence

186
00:09:19,679 --> 00:09:22,600
that a lot of the very experienced
surgeons are,

187
00:09:23,559 --> 00:09:24,840
I don't know that they're
not able to,

188
00:09:24,840 --> 00:09:28,080
but they certainly don't know how
to use the device safely.

189
00:09:29,039 --> 00:09:29,399
Okay?

190
00:09:29,759 --> 00:09:31,519
And to what I said to the
manufacturer,

191
00:09:31,559 --> 00:09:36,600
I said, you really want to be
able to train the surgeons

192
00:09:36,600 --> 00:09:40,639
to use your device safely
according to the IFU.

193
00:09:41,320 --> 00:09:43,320
And what they were concerned
about was

194
00:09:43,480 --> 00:09:46,240
that what I was proposing was
that they were training

195
00:09:46,240 --> 00:09:47,720
surgical skills and I wasn't.

196
00:09:48,279 --> 00:09:50,559
I was just arguing that they
should be training

197
00:09:50,559 --> 00:09:52,240
to use the device safely.

198
00:09:53,399 --> 00:09:55,159
Yeah, I would agree
with that,

199
00:09:55,200 --> 00:09:56,200
with what you just said on,

200
00:09:56,240 --> 00:09:59,200
I mean, this is all just about
doing the stapling itself.

201
00:09:59,200 --> 00:10:00,720
It's not about doing
the procedure.

202
00:10:00,720 --> 00:10:03,120
And I mean, that's what the
stapler is, right?

203
00:10:03,679 --> 00:10:05,559
So yeah, I can see that.

204
00:10:05,759 --> 00:10:07,080
But I would have thought
that would be

205
00:10:07,080 --> 00:10:09,440
that the manufacturer would
take that in

206
00:10:09,480 --> 00:10:11,200
as that is their
responsibility,

207
00:10:11,200 --> 00:10:15,080
is to train them how to use
the device effectively,

208
00:10:15,080 --> 00:10:18,279
which basically means safely when
you can write down to it.

209
00:10:18,559 --> 00:10:19,360
That's correct.

210
00:10:19,639 --> 00:10:21,559
I mean, and a lot of the
problems that we see

211
00:10:21,559 --> 00:10:24,360
with new devices or even
old devices,

212
00:10:25,960 --> 00:10:27,960
the clinicians will blame
the device

213
00:10:28,320 --> 00:10:29,759
but say that if something
was wrong,

214
00:10:29,759 --> 00:10:30,759
it was the device fault.

215
00:10:31,200 --> 00:10:33,879
And I mean, for the cardiovascular
devices,

216
00:10:33,879 --> 00:10:38,240
the implanting, the cardiac
defibrillator.

217
00:10:40,200 --> 00:10:45,879
See, clinicians, they don't follow
the instruction for use.

218
00:10:45,879 --> 00:10:48,960
For example, when you're
suturing the device

219
00:10:49,000 --> 00:10:50,759
into the packet that
you've created,

220
00:10:51,360 --> 00:10:53,720
some clinicians don't use
any sutures at all.

221
00:10:53,720 --> 00:10:57,000
They don't use the sleeve
that covers the leads.

222
00:10:57,279 --> 00:10:59,279
They use the wrong type
of sutures.

223
00:10:59,279 --> 00:11:03,120
They use to dispose or sorry,
sutures that melt away

224
00:11:03,360 --> 00:11:05,080
rather than, you know,

225
00:11:05,600 --> 00:11:05,919
For a minute.

226
00:11:06,279 --> 00:11:08,399
Surely I would have thought
that the manufacturers

227
00:11:08,399 --> 00:11:09,559
would be concerned
about that.

228
00:11:11,279 --> 00:11:12,120
I would think so.

229
00:11:12,399 --> 00:11:15,080
Yeah, but yes, they're
concerned about it.

230
00:11:15,360 --> 00:11:18,759
And their trainers were
telling them

231
00:11:18,759 --> 00:11:20,639
that maybe some of the
clinicians

232
00:11:20,799 --> 00:11:23,000
that were training
the juniors

233
00:11:23,279 --> 00:11:27,399
were not maybe as good as they
originally perceived.

234
00:11:27,639 --> 00:11:29,000
You know, they had a senior
position.

235
00:11:29,080 --> 00:11:32,720
They had a high volume of
procedures per year

236
00:11:32,720 --> 00:11:34,840
according to their CV.

237
00:11:35,080 --> 00:11:37,240
But when you looked at their
performance,

238
00:11:37,240 --> 00:11:38,799
they really didn't
do that well.

239
00:11:39,200 --> 00:11:42,279
And over the last decade, what
we've been publishing,

240
00:11:42,320 --> 00:11:44,440
you know, for the concert
validity of the metrics

241
00:11:44,440 --> 00:11:45,639
for the different devices,

242
00:11:46,000 --> 00:11:47,519
what we've been observing is

243
00:11:47,519 --> 00:11:49,639
that some very senior
clinicians

244
00:11:49,879 --> 00:11:52,360
perform worse than the
worst trainee.

245
00:11:53,960 --> 00:11:55,960
Now that's problematic.

246
00:11:56,840 --> 00:11:58,960
Yeah, no, that's not good.

247
00:11:59,559 --> 00:12:03,480
Yeah, it's, you know, the one
question there is,

248
00:12:03,480 --> 00:12:05,159
you know, are they really
that bad

249
00:12:05,159 --> 00:12:08,840
or is it because they're
not acclimated

250
00:12:08,840 --> 00:12:11,360
to the device and the technology
never have been,

251
00:12:11,480 --> 00:12:13,159
you never learned how
to use it well.

252
00:12:13,600 --> 00:12:14,799
This wasn't on the device.

253
00:12:14,799 --> 00:12:18,200
This was actually video recording
from actual patients.

254
00:12:18,960 --> 00:12:20,159
Okay, just, okay.

255
00:12:20,440 --> 00:12:23,159
Yeah, and the metrics
weren't made by,

256
00:12:23,159 --> 00:12:25,080
us, they were made by
their peers

257
00:12:25,480 --> 00:12:27,080
and for a straightforward
procedure,

258
00:12:27,120 --> 00:12:28,480
not for a complex procedure.

259
00:12:29,120 --> 00:12:31,919
Yeah, yeah, well, these,
as you know,

260
00:12:33,240 --> 00:12:34,559
when you're doing things
like this,

261
00:12:34,559 --> 00:12:35,279
we were talking about,

262
00:12:35,279 --> 00:12:37,639
these are all gonna be observational
metrics, right?

263
00:12:38,159 --> 00:12:40,679
Because it's surgeons watching
other surgeons

264
00:12:40,679 --> 00:12:43,440
so they don't have any other
thing they can do

265
00:12:43,440 --> 00:12:45,799
other than watch and say
what's going on.

266
00:12:45,799 --> 00:12:47,320
Whereas, you know, like
in a simulator,

267
00:12:47,320 --> 00:12:50,639
you could actually measure
the actual forces

268
00:12:50,639 --> 00:12:52,679
and, you know, things
like that,

269
00:12:52,720 --> 00:12:54,960
you know, like tool tissue
interactions and stuff

270
00:12:55,480 --> 00:12:57,360
and turn those into metrics
if you wanted to,

271
00:12:57,360 --> 00:12:59,559
but then nobody would know what
those metrics need to be

272
00:12:59,799 --> 00:13:02,039
because we don't know what that
data is in real life.

273
00:13:02,039 --> 00:13:04,399
So we're kind of stuck doing
the same thing,

274
00:13:04,879 --> 00:13:07,200
just making it fancy and
quantitative

275
00:13:07,200 --> 00:13:10,360
in terms of like tools moving
around and stuff.

276
00:13:10,399 --> 00:13:12,519
You know, all the economy,
emotional,

277
00:13:12,519 --> 00:13:13,639
the other stuff.

278
00:13:13,919 --> 00:13:15,000
We'll get to that
one another.

279
00:13:15,559 --> 00:13:18,919
So when you were building the
simulator in the 1990s,

280
00:13:19,000 --> 00:13:24,399
what was your view on what the
simulator should measure?

281
00:13:25,200 --> 00:13:25,879
And-

282
00:13:26,080 --> 00:13:27,279
Basically everything.

283
00:13:29,559 --> 00:13:31,720
There's only one thing worse
than having no data.

284
00:13:31,919 --> 00:13:32,960
That's having too much data.

285
00:13:32,960 --> 00:13:33,639
It's too much.

286
00:13:35,000 --> 00:13:36,559
So measuring everything.

287
00:13:37,960 --> 00:13:40,519
Well, no, no, no, no, there's
a reason for that.

288
00:13:41,039 --> 00:13:44,159
So when I did the original
VIST,

289
00:13:44,799 --> 00:13:48,120
I did try to do a harness
on it

290
00:13:48,120 --> 00:13:50,559
that would record literally
everything,

291
00:13:51,000 --> 00:13:53,679
like every force, every
torque, you know,

292
00:13:53,679 --> 00:13:56,080
every contact point, like
everything,

293
00:13:56,399 --> 00:13:59,200
because I figured eventually we
were gonna wanna go through

294
00:13:59,200 --> 00:14:00,639
there and try and make
sense of it

295
00:14:00,639 --> 00:14:02,919
in terms of what would
give us measures

296
00:14:02,919 --> 00:14:04,840
that would be kind of
telling you

297
00:14:04,879 --> 00:14:07,840
whether you did it right,
whatever it was.

298
00:14:08,159 --> 00:14:11,320
And so the original intent
was literally to record

299
00:14:11,320 --> 00:14:12,679
as many things as we could

300
00:14:12,960 --> 00:14:14,519
and then go rifling
through it

301
00:14:14,519 --> 00:14:17,919
and try and figure out what
the essential measurements

302
00:14:17,919 --> 00:14:19,320
would be that would tell you

303
00:14:19,320 --> 00:14:20,840
whether you had actually
learned how.

304
00:14:21,320 --> 00:14:24,639
The, so the simulators were
originally supposed

305
00:14:24,639 --> 00:14:26,879
to be designed around
learning goals.

306
00:14:27,159 --> 00:14:29,600
And we wanna be able to have
a way to actually

307
00:14:29,600 --> 00:14:32,440
a certain that you had learned
whatever the goal was.

308
00:14:32,720 --> 00:14:36,039
And so our intent was that we
would have some kind

309
00:14:36,039 --> 00:14:37,799
of measurement of that.

310
00:14:37,840 --> 00:14:39,840
And since we didn't know
what that would be,

311
00:14:40,279 --> 00:14:42,159
we were like, let's just do
built-in suspenders

312
00:14:42,159 --> 00:14:43,639
and record everything and
then we'll figure out

313
00:14:43,639 --> 00:14:45,639
how to pare it down to something
that makes sense.

314
00:14:47,240 --> 00:14:48,879
Well, to some extent,
we did that

315
00:14:48,879 --> 00:14:52,720
for the mechanical thrombectomy
for acute stroke.

316
00:14:53,360 --> 00:14:56,360
I mean, we developed the metrics
from the human procedure

317
00:14:57,279 --> 00:14:58,360
along with Mentice.

318
00:14:58,639 --> 00:15:00,840
And then what we did was,

319
00:15:01,840 --> 00:15:04,279
myself and the three interventional
neuro radiologists,

320
00:15:04,440 --> 00:15:07,399
we imposed the metrics
well, Mentice did.

321
00:15:07,559 --> 00:15:08,720
We told them what to do

322
00:15:08,759 --> 00:15:10,799
and they imposed the metrics
on the system.

323
00:15:11,399 --> 00:15:14,279
And one of the things that
the simulator didn't do

324
00:15:14,519 --> 00:15:17,080
was when you inject contrast
into the brain

325
00:15:17,399 --> 00:15:21,000
and into the brain neurovascular
structure

326
00:15:22,320 --> 00:15:26,039
through a microcatheter, they
were using syringe with air.

327
00:15:26,519 --> 00:15:30,279
And it just plunged straight
down and--

328
00:15:30,440 --> 00:15:31,919
Wrong flow characteristics.

329
00:15:32,039 --> 00:15:32,879
Absolutely.

330
00:15:33,240 --> 00:15:36,039
And the interventional neuro radiologist
stood the ground.

331
00:15:36,039 --> 00:15:37,559
They basically said, that's
not gonna do.

332
00:15:37,840 --> 00:15:39,440
Yeah. And we're not standing
over this.

333
00:15:39,559 --> 00:15:42,960
So Mentice had to develop and
subsequently patent

334
00:15:43,240 --> 00:15:46,600
a new methodology where they
were injecting a fluid

335
00:15:46,840 --> 00:15:50,639
into a small device that
actually was made

336
00:15:50,639 --> 00:15:52,279
the whole training more
realistic.

337
00:15:53,000 --> 00:15:57,200
And I mean, it works when
you insist on it,

338
00:15:57,240 --> 00:15:59,080
but you have to start off
with the metrics.

339
00:15:59,519 --> 00:16:01,600
So you were trying to measure
everything

340
00:16:01,840 --> 00:16:05,720
in the late 1990s,
early 2000s.

341
00:16:06,200 --> 00:16:07,200
What did you end up,

342
00:16:07,279 --> 00:16:09,200
what sort of metrics did you
end up producing?

343
00:16:09,879 --> 00:16:10,480
Or the simulator?

344
00:16:10,480 --> 00:16:16,279
Well, so I really was not the
one that did that.

345
00:16:16,399 --> 00:16:19,120
That was others that decided
what the metrics would be.

346
00:16:19,240 --> 00:16:21,039
I just put the platform
together

347
00:16:21,039 --> 00:16:22,879
so that the information would
be available.

348
00:16:23,360 --> 00:16:25,200
And then other people made
decisions about

349
00:16:25,240 --> 00:16:30,000
what the actual metrics
derived

350
00:16:30,000 --> 00:16:31,480
from that information
would be.

351
00:16:32,120 --> 00:16:35,840
These were clinicians or clinician
engineers or engineers?

352
00:16:37,399 --> 00:16:39,080
Yeah, my experience was that

353
00:16:39,080 --> 00:16:40,320
it was a little bit
of everything.

354
00:16:40,399 --> 00:16:43,679
It was clinicians in concert
with the companies.

355
00:16:44,000 --> 00:16:46,639
The companies usually worked
with some clinicians

356
00:16:46,639 --> 00:16:48,320
and it was usually a
small number.

357
00:16:49,200 --> 00:16:53,759
And whoever their subject
matter experts was

358
00:16:53,799 --> 00:16:56,679
that they were paying to
consult with them.

359
00:16:57,440 --> 00:16:59,200
And then they would basically
for the most part

360
00:16:59,200 --> 00:17:01,399
do whatever their SMEs said.

361
00:17:01,720 --> 00:17:03,559
And all we did was
implement it.

362
00:17:04,079 --> 00:17:06,680
Which usually meant just
pulling out things

363
00:17:06,680 --> 00:17:07,839
that were already there.

364
00:17:08,119 --> 00:17:10,359
It wasn't that often we had
to do something new.

365
00:17:11,359 --> 00:17:13,240
So it was a combination
of stuff

366
00:17:13,240 --> 00:17:14,640
that was already getting
dumped out

367
00:17:14,640 --> 00:17:16,640
or that was already available
in the simulator.

368
00:17:17,039 --> 00:17:19,480
And then turning those into
some kind of metrics

369
00:17:19,480 --> 00:17:21,839
which were typically,
as you know,

370
00:17:22,240 --> 00:17:23,720
they're observational
metrics.

371
00:17:23,720 --> 00:17:25,880
They're nothing exotic for
the most part.

372
00:17:27,400 --> 00:17:27,920
And

373
00:17:28,920 --> 00:17:33,839
I know that you keep an eye on
the simulation developments.

374
00:17:35,000 --> 00:17:38,279
What's your view on the metrics
that currently exist?

375
00:17:38,319 --> 00:17:40,759
Not just in endovascular
simulators or

376
00:17:41,160 --> 00:17:42,640
interventional cardiology
simulators,

377
00:17:42,640 --> 00:17:43,880
but robotics and so on.

378
00:17:45,400 --> 00:17:47,319
They haven't changed a
lot, basically.

379
00:17:48,200 --> 00:17:49,160
They're kind of the
same things

380
00:17:49,160 --> 00:17:50,680
we came up at the very
beginning

381
00:17:50,759 --> 00:17:51,960
when you come right
down to it.

382
00:17:52,319 --> 00:17:53,559
I mean, I think it's
slightly worse.

383
00:17:53,880 --> 00:17:54,440
Yeah, I'm sure there's...

384
00:17:56,000 --> 00:17:57,960
Okay, you're much more in the

385
00:17:57,960 --> 00:18:00,039
fray than I am these
days, frankly.

386
00:18:00,119 --> 00:18:03,839
I was at the ERA conference
in Paris a few months ago,

387
00:18:04,400 --> 00:18:05,880
and I walked around all the

388
00:18:05,880 --> 00:18:07,960
booths to look at the
simulations.

389
00:18:08,680 --> 00:18:11,200
And almost none of them, I
think only one of them,

390
00:18:11,640 --> 00:18:13,799
had any form of metrics
whatsoever.

391
00:18:14,440 --> 00:18:15,079
So it was...

392
00:18:15,960 --> 00:18:16,839
I mean, I was...

393
00:18:17,079 --> 00:18:18,559
To be honest with you, I was
kind of shocked.

394
00:18:19,279 --> 00:18:20,000
And...

395
00:18:20,039 --> 00:18:22,519
Well, were they just using
them as a marketing tool?

396
00:18:22,559 --> 00:18:24,039
I mean, a lot of the early

397
00:18:24,039 --> 00:18:26,240
simulators we did were just
marketing tools.

398
00:18:26,279 --> 00:18:27,720
You know, you stuck them
in a booth

399
00:18:28,039 --> 00:18:30,359
and it was a way to show
off the system,

400
00:18:30,400 --> 00:18:32,440
you know, whatever the system
was you were trying to sell.

401
00:18:32,720 --> 00:18:34,519
And they really didn't care if
there were metrics or not.

402
00:18:34,519 --> 00:18:36,440
It was just to give people
an experience.

403
00:18:36,880 --> 00:18:40,319
Yeah. Well, I also talked to
some of the manufacturers,

404
00:18:40,319 --> 00:18:41,559
the device manufacturers,
and they

405
00:18:41,559 --> 00:18:43,119
were complaining about
the simulators

406
00:18:43,720 --> 00:18:45,680
and the fact that they
weren't able...

407
00:18:45,799 --> 00:18:48,680
Well, some of the manufacturers,
device manufacturers,

408
00:18:48,680 --> 00:18:50,079
didn't want to measure
things.

409
00:18:50,480 --> 00:18:50,799
Because...

410
00:18:51,920 --> 00:18:52,680
Because if they...

411
00:18:52,759 --> 00:18:53,359
They don't want to know.

412
00:18:54,079 --> 00:18:55,319
Correct. They don't
want to know.

413
00:18:55,920 --> 00:18:57,240
Plausible deniability.

414
00:18:57,599 --> 00:18:58,759
Plausible deniability.

415
00:18:59,559 --> 00:19:04,240
And other manufacturers complained
that the simulation

416
00:19:04,240 --> 00:19:06,279
companies weren't producing
the

417
00:19:06,599 --> 00:19:09,599
metrics that they required. And
they said, where do you

418
00:19:09,640 --> 00:19:11,319
expect them to get the
metrics from?

419
00:19:11,599 --> 00:19:15,839
You really have to derive the
metrics from real clinicians

420
00:19:15,839 --> 00:19:19,119
doing real procedures
in real life

421
00:19:19,119 --> 00:19:21,640
situations. And a straightforward
procedure, not a complex

422
00:19:21,640 --> 00:19:23,599
procedure. And when
they go to the

423
00:19:23,599 --> 00:19:27,559
manufacturer, they say, these
are our metrics, our

424
00:19:27,559 --> 00:19:31,319
validated metrics. If you
can build the

425
00:19:31,319 --> 00:19:33,799
metrics into your simulator,
we'll purchase it. If you

426
00:19:33,799 --> 00:19:38,160
can't, we're not. So
I mean, go on.

427
00:19:38,160 --> 00:19:39,480
That would require people
to actually

428
00:19:39,519 --> 00:19:41,279
say that, that that's what
they want to do.

429
00:19:42,759 --> 00:19:44,559
You know, if people
still buy the

430
00:19:44,559 --> 00:19:46,279
simulators, even if their
metrics aren't there,

431
00:19:47,279 --> 00:19:51,000
you know, what am I supposed
to say? Yeah, well, I think

432
00:19:51,000 --> 00:19:52,680
something's got to
give, because

433
00:19:53,200 --> 00:19:57,680
what we're seeing is that the evidence
is fairly clear now.

434
00:19:58,160 --> 00:20:00,119
The intraoperative performance
of the

435
00:20:00,160 --> 00:20:03,000
clinician is going to impact
on the patient outcomes. I

436
00:20:03,000 --> 00:20:07,119
mean, the data is there
now. And

437
00:20:07,119 --> 00:20:12,440
with a new device, it's more likely
that a clinician who's

438
00:20:12,440 --> 00:20:15,440
not trained, you know,
to a level of

439
00:20:15,440 --> 00:20:18,119
proficiency, they're going to
perform less well than

440
00:20:18,119 --> 00:20:19,759
somebody who's actually
been trained to

441
00:20:19,759 --> 00:20:23,799
efficiency. And that's the data
on that's very clear. So I

442
00:20:23,839 --> 00:20:25,920
don't understand why
there isn't,

443
00:20:27,160 --> 00:20:29,359
you know, a drive by the manufacturer,
the device

444
00:20:29,359 --> 00:20:33,000
manufacturers to insist
on metrics being

445
00:20:33,440 --> 00:20:40,440
implemented in the simulator.
I don't know what to tell

446
00:20:40,440 --> 00:20:42,079
you. I mean, maybe they
don't think that

447
00:20:42,079 --> 00:20:45,680
it's going to doesn't make a
difference on selling their

448
00:20:45,680 --> 00:20:49,640
gadgets in any form. It's
just a cost,

449
00:20:49,680 --> 00:20:53,240
to do this. And so if it's not
going to make them sell

450
00:20:53,240 --> 00:20:55,200
more, why do it? I
don't know.

451
00:20:55,880 --> 00:20:59,400
Well, I mean, the other side
of that is there's devices

452
00:20:59,400 --> 00:21:01,440
that have failed in the
marketplace.

453
00:21:01,640 --> 00:21:06,359
And it wasn't an issue with the
device. It was the user of

454
00:21:06,359 --> 00:21:09,599
the device. It's been
held back

455
00:21:10,000 --> 00:21:13,640
and repeated the introduction
of that device into the

456
00:21:13,640 --> 00:21:16,400
market. And I've been
told that it's

457
00:21:16,440 --> 00:21:20,799
hundreds and hundreds of millions
in some cases. Yeah, I

458
00:21:20,839 --> 00:21:22,519
think a lot of times people
don't think

459
00:21:22,519 --> 00:21:25,759
ahead of like think ahead like
that. Or those or those

460
00:21:25,759 --> 00:21:27,920
stories don't get around
well enough,

461
00:21:27,920 --> 00:21:31,279
right to management at the medical
device companies to

462
00:21:31,279 --> 00:21:34,400
realise, really, you know,
you could. Yeah.

463
00:21:36,200 --> 00:21:42,680
Hmm. So in the early 2000s, she
appeared to move away from

464
00:21:43,000 --> 00:21:44,920
simulation and more
into surgical

465
00:21:44,960 --> 00:21:48,799
robotics. What were the major
drivers of that change for

466
00:21:48,799 --> 00:21:52,119
you? So I basically
kind of felt

467
00:21:52,880 --> 00:21:55,160
hemmed in with surgical
simulation.

468
00:21:55,759 --> 00:21:57,599
You were always constrained
by,

469
00:21:58,559 --> 00:22:00,680
there's not that many things
to be simulated,

470
00:22:00,680 --> 00:22:01,880
at least at that point.

471
00:22:02,000 --> 00:22:05,839
And it was always driven by
the device manufacturers,

472
00:22:06,599 --> 00:22:08,440
whatever it was that we were
gonna be doing.

473
00:22:08,720 --> 00:22:10,440
And quite frankly, I kind
of felt like

474
00:22:10,440 --> 00:22:12,119
it was kind of a lousy
business

475
00:22:12,480 --> 00:22:14,160
because you were inherently
capped

476
00:22:14,160 --> 00:22:16,799
as to how many things you were
gonna be able to sell,

477
00:22:16,799 --> 00:22:18,599
how much money you were gonna
be able to make.

478
00:22:18,920 --> 00:22:21,039
Not that I really cared that
much about it,

479
00:22:21,039 --> 00:22:24,240
but it just didn't seem like it
was a major growth field.

480
00:22:24,759 --> 00:22:27,920
And surgical robotics had really
just started to get going.

481
00:22:28,559 --> 00:22:30,519
And it was interesting,

482
00:22:30,559 --> 00:22:32,640
and there was a whole bunch
of different things

483
00:22:32,640 --> 00:22:34,000
to be done there.

484
00:22:34,000 --> 00:22:37,240
So I just decided to move
over there.

485
00:22:37,240 --> 00:22:39,359
And as it turns out that
surgical robotics

486
00:22:39,359 --> 00:22:41,319
isn't all that different from
surgical simulation

487
00:22:41,319 --> 00:22:44,400
in terms of sort of the systems
engineering stuff,

488
00:22:44,400 --> 00:22:45,839
which is more where
I work at it.

489
00:22:45,839 --> 00:22:49,200
So I like working at the
real-time sensing,

490
00:22:49,200 --> 00:22:52,319
real-time computation,
real-time actuation,

491
00:22:52,599 --> 00:22:55,119
and then potentially
real-time graphics,

492
00:22:55,640 --> 00:22:57,279
whether it's stereo
or whatever.

493
00:22:58,000 --> 00:23:00,799
And simulators are actually
very similar to robots

494
00:23:00,839 --> 00:23:03,319
as far as that goes in terms
of all the different parts

495
00:23:03,319 --> 00:23:05,400
having to go together and
work in real-time.

496
00:23:05,400 --> 00:23:09,279
So I basically had a skillset
already done

497
00:23:09,279 --> 00:23:12,119
because of the simulators that
moved over onto robots.

498
00:23:12,480 --> 00:23:15,000
And so I just started working
on the robots

499
00:23:15,640 --> 00:23:18,359
and kind of been going
ever since.

500
00:23:19,079 --> 00:23:20,880
So you agree with Rick
Satavo then

501
00:23:20,880 --> 00:23:24,279
that the VR simulators and
surgical robotics

502
00:23:24,960 --> 00:23:27,440
have an analogous DNA,
if you like,

503
00:23:27,839 --> 00:23:30,559
as they're both information
processing systems?

504
00:23:31,160 --> 00:23:34,039
Yeah, they're both
interesting.

505
00:23:35,200 --> 00:23:38,079
They're both systems where you're
doing real-time sensing

506
00:23:38,079 --> 00:23:40,480
and computation and actuation

507
00:23:40,480 --> 00:23:42,839
and typically graphics
generation.

508
00:23:43,279 --> 00:23:47,200
The simulators don't nearly
have as high a bar to them

509
00:23:47,240 --> 00:23:48,279
as the robots do.

510
00:23:48,319 --> 00:23:50,039
I mean, the robot can actually
hurt somebody.

511
00:23:50,599 --> 00:23:52,640
The simulator can't directly
hurt somebody.

512
00:23:52,640 --> 00:23:55,079
It could indirectly hurt somebody
by teaching them wrong,

513
00:23:55,440 --> 00:23:57,839
but it's not got a direct
immediate effect

514
00:23:57,839 --> 00:23:59,200
of punching a hole
in someone.

515
00:23:59,279 --> 00:24:01,920
So the robots have a much
higher standard

516
00:24:01,920 --> 00:24:03,039
and there's much more that
has to be done

517
00:24:03,039 --> 00:24:04,000
with verification.

518
00:24:04,039 --> 00:24:07,480
In fact, robots do all kinds of
things with verification,

519
00:24:07,480 --> 00:24:09,559
whereas the simulators
basically don't.

520
00:24:10,559 --> 00:24:12,440
What do you mean they do
all sorts of things

521
00:24:12,440 --> 00:24:13,240
with verification?

522
00:24:13,240 --> 00:24:14,519
You mean within the system

523
00:24:14,519 --> 00:24:16,559
or within the management
of the system

524
00:24:16,960 --> 00:24:18,720
by the manufacturer?

525
00:24:19,000 --> 00:24:20,599
When you're developing them,

526
00:24:21,119 --> 00:24:24,640
they put together all kinds of
bench tests for the robots

527
00:24:24,640 --> 00:24:27,119
to make sure that they have
the right forces,

528
00:24:27,200 --> 00:24:30,640
the right velocities, the
right accelerations

529
00:24:30,640 --> 00:24:34,480
that they hold up over time,
your reliability,

530
00:24:35,720 --> 00:24:40,319
that they're stable, that the
control loops are stable,

531
00:24:40,920 --> 00:24:42,119
all kinds of things
like that.

532
00:24:42,119 --> 00:24:44,839
There's just dozens and
dozens of tests

533
00:24:45,240 --> 00:24:48,000
that the device manufacturers
put together

534
00:24:48,000 --> 00:24:51,079
to verify that the
specifications

535
00:24:51,079 --> 00:24:54,880
that they've made for the robot
are what the robot does.

536
00:24:55,519 --> 00:24:57,519
And I, at least in my
experience,

537
00:24:57,519 --> 00:24:58,559
there's little to
none of that

538
00:24:58,559 --> 00:25:00,359
that happens with the
surgical simulators.

539
00:25:00,680 --> 00:25:03,519
It's, if anything, it's
validation,

540
00:25:03,720 --> 00:25:07,119
which I basically don't
like validation.

541
00:25:07,519 --> 00:25:11,559
Verification is like you did
the engineering correctly.

542
00:25:11,559 --> 00:25:13,599
Validation is like, "I
think it's okay."

543
00:25:14,599 --> 00:25:16,839
It's not precise for me.

544
00:25:16,880 --> 00:25:19,680
That's like the end user
just saying,

545
00:25:19,759 --> 00:25:21,000
"Okay, that worked,
I like it."

546
00:25:21,359 --> 00:25:23,240
But they don't know why they
like it or don't like it.

547
00:25:23,640 --> 00:25:25,240
I'm not sure I agree
with that.

548
00:25:25,559 --> 00:25:26,240
I mean, validation-

549
00:25:26,240 --> 00:25:27,559
This is from an engineering
point.

550
00:25:28,200 --> 00:25:29,519
This has been an engineering
point of view.

551
00:25:30,240 --> 00:25:34,160
Clinically, I mean, it's
not my methodology.

552
00:25:34,559 --> 00:25:37,680
The methodology was agreed
in the 1970s

553
00:25:37,920 --> 00:25:40,960
by all the education and
training establishments

554
00:25:40,960 --> 00:25:43,200
in terms of how you validate,

555
00:25:43,400 --> 00:25:44,960
whether something works or
it doesn't work.

556
00:25:45,279 --> 00:25:47,960
Well, okay, so from an engineering
point of view,

557
00:25:47,960 --> 00:25:52,559
validation is like, "Yeah, okay,
it does something useful.

558
00:25:53,000 --> 00:25:55,000
They're actually learning
or whatever it is."

559
00:25:55,759 --> 00:25:58,519
But I find that to
be imprecise.

560
00:25:59,720 --> 00:26:01,079
From an engineering
point of view,

561
00:26:01,079 --> 00:26:03,200
I'm much more interested
in verification.

562
00:26:03,519 --> 00:26:05,319
I guess the way I look
at it from a,

563
00:26:05,359 --> 00:26:07,440
like from a surgical robot
point of view,

564
00:26:08,359 --> 00:26:10,000
verification and validation
is,

565
00:26:10,039 --> 00:26:12,519
verification is I built what
I said I was gonna build,

566
00:26:12,839 --> 00:26:14,960
and validation is I built
the right thing.

567
00:26:15,839 --> 00:26:19,759
I consider that much
less precise

568
00:26:19,759 --> 00:26:20,960
from an engineering
point of view.

569
00:26:21,599 --> 00:26:23,599
Yeah, but it isn't
engineering.

570
00:26:23,599 --> 00:26:26,519
And I mean, the ultimate test
of the validation

571
00:26:26,519 --> 00:26:28,319
for a robot or a simulator
is,

572
00:26:28,519 --> 00:26:32,680
does the robot do what it's
supposed to do clinically?

573
00:26:33,440 --> 00:26:36,559
For somebody that can it
do the procedure,

574
00:26:36,559 --> 00:26:38,119
can it do the procedure
safely?

575
00:26:38,200 --> 00:26:42,480
And that's not the engineer's
realm, basically.

576
00:26:42,680 --> 00:26:43,640
That's my point of view.

577
00:26:43,880 --> 00:26:46,720
That's for the clinicians
and others,

578
00:26:47,000 --> 00:26:48,480
but the engineer is
to verify.

579
00:26:49,400 --> 00:26:55,799
Yeah, so in the 1990s, laparoscopic
surgery developed,

580
00:26:56,440 --> 00:27:01,640
they extensively investigated
this whole surgical tracking

581
00:27:01,640 --> 00:27:04,920
and path and path length and
motion and so on.

582
00:27:05,480 --> 00:27:08,119
And it really didn't
go anywhere.

583
00:27:08,559 --> 00:27:10,920
And surgical robotics seems
to be heading down

584
00:27:10,960 --> 00:27:15,359
the same road as an approach
to the assessment

585
00:27:15,599 --> 00:27:18,680
of operative performance.

586
00:27:19,079 --> 00:27:19,519
Sure.

587
00:27:20,240 --> 00:27:22,240
Oh, I think there's two
parts to this.

588
00:27:23,279 --> 00:27:25,160
It's basically the only thing
people can think of

589
00:27:25,160 --> 00:27:26,079
for the most part.

590
00:27:26,880 --> 00:27:28,480
And it's gonna get
turbocharged

591
00:27:28,519 --> 00:27:30,319
because of machine learning
in AI.

592
00:27:30,519 --> 00:27:30,920
Yeah.

593
00:27:31,160 --> 00:27:32,599
It's gonna be on steroids

594
00:27:32,599 --> 00:27:33,960
because they're gonna be
looking for patterns

595
00:27:33,960 --> 00:27:36,680
in everything with the data
they collect.

596
00:27:36,680 --> 00:27:37,839
I have a feeling.

597
00:27:38,240 --> 00:27:39,319
Yeah, but I mean,

598
00:27:39,359 --> 00:27:44,000
I mean, it's like looking
for a target

599
00:27:44,000 --> 00:27:46,599
and a very thick fog.

600
00:27:47,559 --> 00:27:49,279
I mean, the whole thing
about science

601
00:27:49,279 --> 00:27:52,079
is that you're supposed
to hypothesise

602
00:27:52,400 --> 00:27:53,359
what it is that you're
looking.

603
00:27:53,720 --> 00:27:57,440
I mean, the tracking stuff,

604
00:27:58,440 --> 00:28:01,799
it seems to be like going on
a fishing expedition

605
00:28:02,000 --> 00:28:02,880
without any--

606
00:28:02,920 --> 00:28:03,839
It is, yep.

607
00:28:04,640 --> 00:28:07,279
I mean, and do you think it's
gonna go anywhere?

608
00:28:08,480 --> 00:28:11,759
I'm sure they'll find
some things.

609
00:28:11,880 --> 00:28:15,079
I mean, this is the thing
with AI coming at it.

610
00:28:15,079 --> 00:28:18,200
It's that it's able to
do so much more

611
00:28:18,680 --> 00:28:20,079
than we could in the past.

612
00:28:22,880 --> 00:28:25,759
But I mean, you really wanna
have a theory of operation

613
00:28:26,200 --> 00:28:29,920
behind things of what is it
we're trying to achieve

614
00:28:29,920 --> 00:28:31,680
and how is it you're
going about

615
00:28:32,680 --> 00:28:34,519
making this thing attain
whatever it is

616
00:28:34,519 --> 00:28:35,480
it's going to attain.

617
00:28:36,039 --> 00:28:41,519
So I would think that you
basically wanna,

618
00:28:42,400 --> 00:28:44,480
you have to have some
kind of concept

619
00:28:44,480 --> 00:28:46,279
that drives why you do
what you do,

620
00:28:46,279 --> 00:28:48,720
which is why you do the
things you do

621
00:28:48,720 --> 00:28:49,960
where you break procedures
down

622
00:28:49,960 --> 00:28:51,839
so you understand that there's
learning goals

623
00:28:52,319 --> 00:28:53,799
and based on those learning
goals

624
00:28:53,799 --> 00:28:57,799
that there's a scaffolding of
skills that support those.

625
00:28:57,799 --> 00:28:58,559
And then from that,

626
00:28:58,559 --> 00:29:00,200
you'd figure out how to
measure things

627
00:29:00,480 --> 00:29:02,559
because those are the things
that support

628
00:29:02,599 --> 00:29:04,799
your understanding of whether
you would obtain

629
00:29:04,799 --> 00:29:05,720
those skills or not.

630
00:29:06,079 --> 00:29:09,799
Yeah, so what strikes me is
that what laparoscopic did

631
00:29:09,799 --> 00:29:12,200
and what robotic surgery
is doing now

632
00:29:12,200 --> 00:29:15,920
and interventional cardiology and
the fluoroscopically,

633
00:29:15,920 --> 00:29:19,720
they measure process and
they take that

634
00:29:19,720 --> 00:29:21,480
as some sort of indication
of skills.

635
00:29:22,079 --> 00:29:23,920
If somebody does lots
of procedures,

636
00:29:24,160 --> 00:29:26,279
they're gonna be better than
somebody that does less.

637
00:29:26,480 --> 00:29:28,200
If you're smooth with
the instrument

638
00:29:28,200 --> 00:29:29,680
or you have a shorter
path length,

639
00:29:30,039 --> 00:29:31,440
you're obviously more skilled

640
00:29:31,440 --> 00:29:34,519
rather than you just jump
complete phases

641
00:29:34,519 --> 00:29:36,160
or steps in the procedure.

642
00:29:36,640 --> 00:29:40,319
And this process seems to be,

643
00:29:40,319 --> 00:29:41,759
I mean, like they're
hanging on to,

644
00:29:41,759 --> 00:29:43,640
this is a strategy for
measurement.

645
00:29:43,880 --> 00:29:47,079
Yeah, well, so if it was
a perfect world,

646
00:29:47,079 --> 00:29:48,799
I'll just use suturing
as an example,

647
00:29:48,799 --> 00:29:50,519
because that's the thing
I tried years ago.

648
00:29:51,480 --> 00:29:52,640
If it was a perfect world,

649
00:29:52,640 --> 00:29:56,039
we'd actually know exactly how
much tissue apposition,

650
00:29:56,039 --> 00:29:59,079
how much of the tissue faces
need to come be in contact

651
00:29:59,400 --> 00:30:01,160
and how much pressure there
needs to be,

652
00:30:02,039 --> 00:30:04,039
and how well distributed
it is.

653
00:30:04,279 --> 00:30:05,880
And that would actually
be the measure

654
00:30:05,880 --> 00:30:07,519
of how well you did suturing.

655
00:30:08,079 --> 00:30:10,119
It wouldn't be an indirect
measure of like,

656
00:30:10,279 --> 00:30:14,000
did you do it smoothly,
quickly, decisively,

657
00:30:14,319 --> 00:30:15,000
all those things,

658
00:30:15,000 --> 00:30:18,119
because that's not really the
measure of the outcome.

659
00:30:18,640 --> 00:30:20,359
We know that the whole idea
with suturing

660
00:30:20,359 --> 00:30:21,559
is you're trying to
get the tissue

661
00:30:21,559 --> 00:30:23,359
to grow back together
and heal, right?

662
00:30:23,839 --> 00:30:25,799
So you'd measure it based
on the metrics

663
00:30:25,799 --> 00:30:27,440
that are reflective
of achieving

664
00:30:27,440 --> 00:30:31,759
the ineffective environment
for tissue regrowth.

665
00:30:32,359 --> 00:30:34,039
But we don't even know
what those are.

666
00:30:35,359 --> 00:30:37,640
Back in the day, at least when
I was working on it,

667
00:30:37,799 --> 00:30:39,599
that wasn't published
information.

668
00:30:39,599 --> 00:30:41,880
Nobody had sat down and
built like a jig

669
00:30:42,240 --> 00:30:44,640
and tried all kinds of different
amounts of pressure

670
00:30:44,640 --> 00:30:48,039
and distribution of the
sutures along it

671
00:30:48,240 --> 00:30:52,119
and the likes to see how well
living tissue would regrow.

672
00:30:52,480 --> 00:30:55,079
So like, nobody knows as
far as I know.

673
00:30:55,720 --> 00:30:57,519
Yeah, well, I mean, I've
worked quite a bit

674
00:30:57,519 --> 00:30:58,640
with suturing as well,

675
00:30:58,880 --> 00:31:00,440
and I haven't seen once
a milliter

676
00:31:00,920 --> 00:31:03,799
that can reasonably, even
reasonably,

677
00:31:04,119 --> 00:31:09,000
approximate the instrument
thread, needle, suture,

678
00:31:09,039 --> 00:31:10,519
or tissue dynamics.

679
00:31:10,799 --> 00:31:12,680
And so what I've reverted to

680
00:31:12,960 --> 00:31:16,079
is actually getting individuals
to suture on tissue

681
00:31:16,640 --> 00:31:18,160
and then a surrogate
measures,

682
00:31:18,440 --> 00:31:21,200
like how well is the tissue
approximated,

683
00:31:21,440 --> 00:31:23,359
did the surgeon drive
the needle

684
00:31:23,559 --> 00:31:25,400
through the tissue
atraumatically,

685
00:31:26,039 --> 00:31:28,440
do their knots hold when
they're supposed to?

686
00:31:29,079 --> 00:31:30,240
Is the tissue pucker?

687
00:31:30,680 --> 00:31:33,880
And those surrogate measures seem
to work reasonably well,

688
00:31:34,160 --> 00:31:37,559
but why hasn't somebody
actually worked out

689
00:31:37,799 --> 00:31:43,319
the actual physics of suturing
and thread dynamics?

690
00:31:44,480 --> 00:31:47,440
Well, the suture and the
tissue, it's not easy.

691
00:31:49,000 --> 00:31:53,359
It's a high computation
activity

692
00:31:53,359 --> 00:31:56,079
and there frankly aren't
that many people

693
00:31:56,079 --> 00:31:57,160
that work in this field.

694
00:31:58,160 --> 00:32:01,039
If this is, so like look
at what's happening

695
00:32:01,039 --> 00:32:01,920
with AI right now.

696
00:32:02,759 --> 00:32:07,119
The entire planet has decided
that having physics,

697
00:32:07,640 --> 00:32:09,440
what they're calling physical
intelligence now

698
00:32:09,480 --> 00:32:12,000
is sort of one of the next
big things for AI.

699
00:32:12,480 --> 00:32:15,480
So there's like thousands, tens
of thousands of people

700
00:32:15,480 --> 00:32:17,880
doing physics models
now for robots,

701
00:32:18,119 --> 00:32:19,880
for humanoid robots and
everything else.

702
00:32:20,279 --> 00:32:23,160
And you're gonna see a tremendous
amount of improvement

703
00:32:23,160 --> 00:32:25,480
in the physics simulation
abilities.

704
00:32:25,759 --> 00:32:27,960
You just don't have that with
surgical simulation.

705
00:32:28,599 --> 00:32:30,240
It's not like the whole
planet's decided

706
00:32:30,240 --> 00:32:31,839
this is how you're gonna
become a billionaire

707
00:32:32,319 --> 00:32:34,880
is by getting physical
intelligence

708
00:32:34,880 --> 00:32:36,240
combined with humanoid
robots,

709
00:32:36,240 --> 00:32:38,480
which is what a whole lot of
people have decided now.

710
00:32:39,079 --> 00:32:40,799
So you're seeing a huge
amount of improvement.

711
00:32:40,880 --> 00:32:42,839
You're just not gonna see
that in simulation,

712
00:32:42,839 --> 00:32:44,039
in surgical simulation.

713
00:32:44,599 --> 00:32:46,000
It's just not, like I said,

714
00:32:46,160 --> 00:32:48,039
that's the reason I got out
of it a long time ago.

715
00:32:48,039 --> 00:32:49,160
It's just not a big market.

716
00:32:49,640 --> 00:32:51,440
So there's not a lot of
drive to do it.

717
00:32:51,839 --> 00:32:53,759
What's gonna happen with
surgical simulation

718
00:32:53,799 --> 00:32:56,759
is they're gonna get the
indirect benefits

719
00:32:56,759 --> 00:32:59,440
of all the work that's happening
for robotics right now.

720
00:33:00,440 --> 00:33:03,599
And you have all these things
that people are doing

721
00:33:03,599 --> 00:33:06,839
where they're using AI combined
with physics models now,

722
00:33:07,160 --> 00:33:11,640
so that you can use a reduced
accuracy physics model

723
00:33:11,640 --> 00:33:14,160
and you use AI to fill
in the blanks.

724
00:33:14,839 --> 00:33:16,559
I think you'll probably see
that showing up

725
00:33:16,559 --> 00:33:17,759
in surgical simulation,

726
00:33:18,039 --> 00:33:19,720
but it's gonna come from
other fields.

727
00:33:19,720 --> 00:33:21,599
It's not gonna come
from surgical.

728
00:33:21,599 --> 00:33:23,640
It's not gonna be developed
in surgical simulation.

729
00:33:23,680 --> 00:33:25,079
It's gonna be developed
in other fields

730
00:33:25,319 --> 00:33:27,119
and adopted in basically.

731
00:33:27,839 --> 00:33:31,880
But if the measurements
of whatever task

732
00:33:31,880 --> 00:33:33,599
you're at let's stick
with suturing,

733
00:33:34,000 --> 00:33:38,759
if the measurements that you have
are measures of process

734
00:33:39,079 --> 00:33:41,279
rather than actual measures
of skill,

735
00:33:41,799 --> 00:33:44,759
I'm not too sure how the AI system's
gonna improve that.

736
00:33:44,920 --> 00:33:47,920
No, yeah, no, no, yeah,
no, that doesn't,

737
00:33:48,559 --> 00:33:50,160
that's such an indirect
measure.

738
00:33:50,160 --> 00:33:52,599
I don't see how that really
is a good reflection

739
00:33:52,599 --> 00:33:55,079
of mastery, if you will.

740
00:33:55,599 --> 00:33:58,279
Yeah, I mean, so you're still
going to need

741
00:33:58,279 --> 00:34:00,359
the performance metrics
which are derived

742
00:34:00,480 --> 00:34:02,720
from experienced clinicians

743
00:34:03,000 --> 00:34:08,280
and to actually teach an AI
system to how to assess.

744
00:34:08,760 --> 00:34:11,599
And I don't see anybody doing
that in the surgical domain.

745
00:34:12,039 --> 00:34:15,440
And I don't know whether they're
actually doing it

746
00:34:15,480 --> 00:34:20,599
in other domains like famous
or infamous, whichever.

747
00:34:21,159 --> 00:34:23,840
Car manufacturer said that
surgeons would be replaced

748
00:34:23,840 --> 00:34:26,519
in what I can remember, was it
two years or three years?

749
00:34:28,559 --> 00:34:30,239
Oh, him, yeah, right.

750
00:34:30,880 --> 00:34:34,280
So what is happening now in
humanoid robotics

751
00:34:34,280 --> 00:34:36,840
is they're trying to teach
all these humanoids.

752
00:34:36,960 --> 00:34:38,760
So they're measuring all
kinds of things

753
00:34:38,760 --> 00:34:40,679
about everyday activities.

754
00:34:41,880 --> 00:34:44,719
So it'll be interesting to
see if some of that stuff

755
00:34:44,719 --> 00:34:49,920
actually maybe becomes learning
points relative

756
00:34:49,960 --> 00:34:51,880
to things that could
be moved over

757
00:34:51,880 --> 00:34:55,239
into surgical simulation,
just the process

758
00:34:55,239 --> 00:34:57,480
by which they're doing it,
not the details of it.

759
00:34:57,719 --> 00:34:58,760
Yeah, but you see--

760
00:34:59,000 --> 00:35:01,320
You are trying to quantify things
like crazy right now

761
00:35:01,920 --> 00:35:04,079
in everyday activities, so.

762
00:35:04,119 --> 00:35:08,039
Yeah, but what I see a lot
of the AI people do

763
00:35:08,280 --> 00:35:13,440
is they're loading data into
the AI system.

764
00:35:13,800 --> 00:35:16,800
And there's no quality, little
or not very good

765
00:35:16,800 --> 00:35:18,519
quality assurance of the data

766
00:35:18,840 --> 00:35:20,760
that the AI system's
learning on.

767
00:35:21,400 --> 00:35:26,280
And the individuals that are
coding the tasks,

768
00:35:26,519 --> 00:35:30,159
I mean, they don't really
know whether the task

769
00:35:30,239 --> 00:35:30,840
could or not.

770
00:35:31,199 --> 00:35:38,480
I mean, but that has to impede
the development of AI

771
00:35:38,519 --> 00:35:41,400
in terms of the assessment
of performance.

772
00:35:43,679 --> 00:35:45,679
Well, in surgery, certainly
it is.

773
00:35:45,880 --> 00:35:48,119
I mean, if you're looking
at all these

774
00:35:48,239 --> 00:35:50,039
humanoid robots desperately
trying

775
00:35:50,039 --> 00:35:53,280
to do everyday activities,
I think the thing

776
00:35:53,280 --> 00:35:55,880
that's happening there is there's
so many people doing it

777
00:35:55,920 --> 00:35:57,760
that it's an aggregate
result.

778
00:35:59,119 --> 00:36:02,440
There's just so many, there's
like 100 humanoid

779
00:36:02,440 --> 00:36:04,480
robotics companies in
China alone.

780
00:36:05,719 --> 00:36:07,800
So you get enough of
them doing it

781
00:36:07,920 --> 00:36:09,800
and they're all piling on
top of each other.

782
00:36:10,119 --> 00:36:11,920
And no matter how much
they're screwing up,

783
00:36:11,960 --> 00:36:14,280
there's enough of them doing
the right thing

784
00:36:14,559 --> 00:36:15,760
that it accumulates.

785
00:36:15,960 --> 00:36:19,000
And you just don't have that effect
with surgery simulation

786
00:36:19,360 --> 00:36:21,639
because there's just not that
many people active in it

787
00:36:21,679 --> 00:36:22,719
when you come right
down to it.

788
00:36:23,920 --> 00:36:27,440
One of the things that has struck
me across the decades

789
00:36:27,599 --> 00:36:29,599
where we've been hanging
out together

790
00:36:29,920 --> 00:36:33,800
is everybody's looking
for an easy way

791
00:36:34,119 --> 00:36:36,440
to actually do something
that's quite difficult.

792
00:36:37,159 --> 00:36:42,599
And science, I mean, doing good
science, it's not easy.

793
00:36:43,079 --> 00:36:47,320
You know, I mean, and I mean,
one of the things

794
00:36:47,320 --> 00:36:50,559
that I'm shocked at is the
use of Likert scales

795
00:36:51,039 --> 00:36:52,320
for the assessment of
performance.

796
00:36:52,559 --> 00:36:54,679
Because again, it's a measure
of process.

797
00:36:55,519 --> 00:36:56,400
Yeah, yeah.

798
00:36:56,960 --> 00:36:59,760
I mean, in robotics, I'm
sorry, go ahead.

799
00:37:00,480 --> 00:37:02,480
No, I was gonna say all those
hand-wavy things, right?

800
00:37:02,719 --> 00:37:04,480
All those hand-wavy. One
to five, right?

801
00:37:05,280 --> 00:37:05,880
One to five.

802
00:37:06,360 --> 00:37:09,239
I mean, my fellows,
well, they

803
00:37:09,239 --> 00:37:10,400
know my views on it,
don't they?

804
00:37:10,639 --> 00:37:14,639
And they initially think that
Tony's got a thing

805
00:37:14,639 --> 00:37:17,679
about Likert scales, and they
very quickly learn,

806
00:37:17,920 --> 00:37:21,239
okay, for the binary metric-based
assessment,

807
00:37:21,239 --> 00:37:23,440
you've got to watch the entire
video and score it.

808
00:37:23,800 --> 00:37:26,760
And you have to learn how
to score it reliably

809
00:37:26,760 --> 00:37:28,960
with somebody else that
you're paired with.

810
00:37:29,280 --> 00:37:32,000
With the Likert scales, you
can watch it, you know,

811
00:37:32,000 --> 00:37:33,800
about two to five minutes
of the video,

812
00:37:34,079 --> 00:37:35,719
and you can score it
in one to five

813
00:37:35,719 --> 00:37:37,159
on the different constructs.

814
00:37:37,400 --> 00:37:40,119
I mean, that's not measurement,
so it's not.

815
00:37:40,519 --> 00:37:41,320
Yep, yep.

816
00:37:42,039 --> 00:37:43,639
So what's your current view

817
00:37:44,360 --> 00:37:47,320
on the status of surgical
robotic simulation

818
00:37:47,599 --> 00:37:50,000
and the metrics used to
assess performance?

819
00:37:50,320 --> 00:37:52,239
I mean, you've worked on
these or you've--

820
00:37:52,800 --> 00:37:54,639
Yeah, I've been around
them, yeah.

821
00:37:54,639 --> 00:37:55,440
You've been around them?

822
00:37:57,159 --> 00:37:58,880
They could be better.

823
00:37:59,679 --> 00:38:04,360
I mean, you know, even doing
the simple things

824
00:38:04,360 --> 00:38:07,000
like ring exercises and
stuff like that,

825
00:38:07,719 --> 00:38:10,679
they still haven't verified that
the performance of the,

826
00:38:11,360 --> 00:38:13,280
you know, like the
exact forces

827
00:38:13,280 --> 00:38:16,199
that you grasp the rings with
and the exact forces

828
00:38:16,199 --> 00:38:17,840
when you pull on the
stretchy ones,

829
00:38:18,079 --> 00:38:21,239
that those actually reflect
what happens in real life.

830
00:38:22,119 --> 00:38:24,800
So it's all based on tweaking
and tuning

831
00:38:25,880 --> 00:38:29,000
to make it so that it's at
least objectionable

832
00:38:29,000 --> 00:38:30,440
to the largest number
of people,

833
00:38:31,480 --> 00:38:33,280
but that doesn't mean that
it's correct.

834
00:38:33,760 --> 00:38:35,800
And I don't think it's ever,

835
00:38:36,360 --> 00:38:38,159
unless they change how
they do things,

836
00:38:38,159 --> 00:38:39,159
it's probably never gonna be,

837
00:38:39,280 --> 00:38:40,960
you're never gonna know
how correct it is

838
00:38:40,960 --> 00:38:42,039
when you come right
down to it.

839
00:38:42,800 --> 00:38:45,599
But surely that should
be of concern

840
00:38:45,599 --> 00:38:47,599
given that you said earlier

841
00:38:47,920 --> 00:38:50,920
that the robot is something
that can actually hurt

842
00:38:51,000 --> 00:38:53,360
a patient relatively easily

843
00:38:53,360 --> 00:38:56,199
if the surgeon doesn't
actually know

844
00:38:56,199 --> 00:38:58,039
how to use it properly
and safely.

845
00:38:58,960 --> 00:39:02,159
Yeah, well, I'm not disagreeing
with you on that.

846
00:39:02,199 --> 00:39:04,840
I mean, but the manufacturer's
goal

847
00:39:04,880 --> 00:39:10,480
is to verify that it won't unintentionally
hurt somebody.

848
00:39:10,880 --> 00:39:13,760
If it intentionally hurts somebody
because of the operator,

849
00:39:13,760 --> 00:39:15,960
that's kind of not their
problem,

850
00:39:16,000 --> 00:39:17,599
is there kind of their
point of view?

851
00:39:18,159 --> 00:39:20,360
Unless of course it happened
so many times

852
00:39:20,360 --> 00:39:21,800
because of something
that's clearly

853
00:39:22,159 --> 00:39:23,840
a human user interface issue,

854
00:39:24,239 --> 00:39:25,800
then that does become
their problem.

855
00:39:27,280 --> 00:39:30,039
And so whose responsibility
is then

856
00:39:30,360 --> 00:39:33,199
for the clinician to
be trained

857
00:39:33,239 --> 00:39:38,840
to a level that allows them
to use the device safely?

858
00:39:40,360 --> 00:39:43,639
Well, I would assume
it would be

859
00:39:43,639 --> 00:39:44,920
the professional societies.

860
00:39:45,000 --> 00:39:47,360
I mean, the ones that are
actually saying

861
00:39:47,360 --> 00:39:48,840
that yes, you are a surgeon.

862
00:39:49,559 --> 00:39:51,079
And if not the societies,

863
00:39:51,079 --> 00:39:54,039
it's the hospitals themselves
giving privileges.

864
00:39:55,360 --> 00:39:57,880
So then that would be sort
of the next level up

865
00:39:57,880 --> 00:40:00,440
because the privilege
granting

866
00:40:00,519 --> 00:40:02,239
is at a whole different level

867
00:40:02,360 --> 00:40:05,239
than the device manufacturers,
right?

868
00:40:05,360 --> 00:40:06,719
The device manufacturer
is just saying,

869
00:40:06,719 --> 00:40:08,440
I taught you how to
use the device

870
00:40:08,719 --> 00:40:12,400
in a way that we certify that
it can be used.

871
00:40:13,159 --> 00:40:15,039
But the granting of privileges
is saying,

872
00:40:15,039 --> 00:40:17,039
you know how to use this
gadget to actually

873
00:40:17,559 --> 00:40:20,760
get a clinical result and
we're gonna take,

874
00:40:21,360 --> 00:40:24,280
we end up having liability
for it

875
00:40:24,280 --> 00:40:25,800
if you don't essentially,
right?

876
00:40:27,280 --> 00:40:30,119
But in the US, you have
a much better,

877
00:40:30,320 --> 00:40:31,559
I'm okay, it's problematic,

878
00:40:31,559 --> 00:40:33,320
but you have a better
privileging system

879
00:40:33,320 --> 00:40:35,719
than we have over in Europe.

880
00:40:36,400 --> 00:40:37,800
I mean, once you're
qualified,

881
00:40:37,840 --> 00:40:40,519
I mean, I was at a PhD viva
by last week

882
00:40:40,519 --> 00:40:42,599
and one of the examiners
was saying

883
00:40:42,599 --> 00:40:45,039
that he works in Sweden.

884
00:40:45,599 --> 00:40:49,840
And once he's qualified as
a surgical specialty,

885
00:40:50,039 --> 00:40:51,400
he can do almost anything.

886
00:40:51,840 --> 00:40:53,599
He also works in New York.

887
00:40:54,159 --> 00:40:58,679
And he said each procedure
that he wants to perform,

888
00:40:58,960 --> 00:41:00,800
he has to be privileged
for it.

889
00:41:01,039 --> 00:41:03,400
That seems like a better
system than here.

890
00:41:03,920 --> 00:41:06,039
Yeah, yeah, no, I would
go with that,

891
00:41:06,039 --> 00:41:07,480
but I think that's sort
of reflective

892
00:41:07,480 --> 00:41:09,679
of sort of the old school,
you know,

893
00:41:09,920 --> 00:41:12,079
doctors are geniuses
kind of thing

894
00:41:12,159 --> 00:41:13,639
from the old days, right?

895
00:41:13,719 --> 00:41:14,920
So, yeah, it's not--

896
00:41:15,079 --> 00:41:17,159
You mean in Europe or the US?

897
00:41:19,039 --> 00:41:21,760
Well, in Europe in that case,
I mean, I don't know if the

898
00:41:22,360 --> 00:41:24,440
experience you said about in
New York, if that's uniform

899
00:41:24,440 --> 00:41:26,920
across the US, I don't really
follow that.

900
00:41:27,360 --> 00:41:29,079
It would be nice if it
was. I suspect

901
00:41:29,079 --> 00:41:31,800
it's not. It's probably
highly variable.

902
00:41:31,880 --> 00:41:35,000
I can only talk about, you
know, I've worked in Yale

903
00:41:35,000 --> 00:41:36,880
and Emory and it was the case

904
00:41:36,880 --> 00:41:38,639
in both of those like 20
odd years ago.

905
00:41:38,639 --> 00:41:39,079
That's good.

906
00:41:39,440 --> 00:41:42,159
But one of the problems, the Chief
of Cardiology, there was

907
00:41:42,159 --> 00:41:43,679
a credentialing going
on for carotid

908
00:41:43,679 --> 00:41:46,079
artery stenting with embolic
protection

909
00:41:46,360 --> 00:41:48,559
and he was saying that he
had to go to one of

910
00:41:48,559 --> 00:41:51,280
the credentialing committees
and I said,

911
00:41:51,639 --> 00:41:53,719
"That sounds, you know,
pretty powerful."

912
00:41:53,719 --> 00:41:54,679
And he said, "No,
not really."

913
00:41:55,079 --> 00:41:59,000
He says the credentialing system
in a hospital is based on

914
00:41:59,000 --> 00:42:00,480
the person who has the lowest

915
00:42:00,480 --> 00:42:03,480
experience in that procedure
on the committee.

916
00:42:04,519 --> 00:42:07,400
So there was again no
objective for it.

917
00:42:08,480 --> 00:42:11,079
Nobody wants to have any kind
of social admittance of

918
00:42:11,079 --> 00:42:13,559
social whatever that they
don't know what's

919
00:42:13,559 --> 00:42:14,039
going on.

920
00:42:14,039 --> 00:42:14,599
Yeah, okay.

921
00:42:14,760 --> 00:42:19,639
I mean, I think that at some point
there needs to be some

922
00:42:19,679 --> 00:42:22,039
objective assessment
of performance,

923
00:42:22,320 --> 00:42:24,400
either at the end of
a course or a

924
00:42:24,440 --> 00:42:28,880
verification, you know, at
a hospital level.

925
00:42:29,159 --> 00:42:32,039
Because one of the things we're
learning is a lot of the

926
00:42:32,039 --> 00:42:34,039
metrics in the past were used

927
00:42:34,320 --> 00:42:35,679
for the assessment of
performance,

928
00:42:35,679 --> 00:42:37,280
like how many procedures
did you do?

929
00:42:37,320 --> 00:42:38,400
Did you do a fellowship?

930
00:42:38,639 --> 00:42:39,920
How long was your fellowship?

931
00:42:40,159 --> 00:42:42,119
These were the performance
metrics that

932
00:42:42,119 --> 00:42:44,480
were used to say, "Okay, that
person's safe to

933
00:42:44,480 --> 00:42:45,360
do the procedure."

934
00:42:45,719 --> 00:42:48,360
But one of the things that we've
learned over the last few

935
00:42:48,360 --> 00:42:49,719
decades since the
introduction

936
00:42:49,719 --> 00:42:51,920
of the laparoscopic surgery,
to some

937
00:42:51,920 --> 00:42:54,800
extent, maybe even to a large
extent, those sorts

938
00:42:54,800 --> 00:42:56,119
of metrics mean nothing.

939
00:42:56,119 --> 00:42:57,519
They're measures of
process again.

940
00:42:58,320 --> 00:43:01,199
Yeah, well, they're social proof,
if anything, when you

941
00:43:01,239 --> 00:43:02,559
come right down to it, right?

942
00:43:03,079 --> 00:43:05,159
I know this person and I
think they're okay.

943
00:43:05,199 --> 00:43:06,719
And that's kind of the
end of it, right?

944
00:43:07,840 --> 00:43:09,920
Yeah, and do you think
that's okay?

945
00:43:11,159 --> 00:43:12,320
No, of course not.

946
00:43:13,239 --> 00:43:19,400
I mean, we fall back to what Rick
Satava and others went on

947
00:43:19,400 --> 00:43:20,679
and on and on about
years ago,

948
00:43:20,679 --> 00:43:23,239
about the way that the pilots
are done, right?

949
00:43:24,079 --> 00:43:27,000
Mandatory training and mandatory
testing, right?

950
00:43:27,079 --> 00:43:27,960
Yes, yes.

951
00:43:28,239 --> 00:43:29,920
And that's that end of story.

952
00:43:30,119 --> 00:43:34,280
And those simulators, just to
throw this in for reference,

953
00:43:34,480 --> 00:43:36,360
those simulators are
verified.

954
00:43:37,079 --> 00:43:40,800
Those simulators are verified
big time, that they give the

955
00:43:40,800 --> 00:43:42,239
right movements, they
give the

956
00:43:42,280 --> 00:43:46,480
right forces, you know, everything
on those simulators is

957
00:43:46,480 --> 00:43:47,920
reflective of what happens in

958
00:43:47,920 --> 00:43:48,559
real life.

959
00:43:50,920 --> 00:43:55,440
But they cost like tens or hundreds
of millions to develop?

960
00:43:55,599 --> 00:43:56,199
Absolutely.

961
00:43:56,719 --> 00:43:59,679
And so why do you think that
the governments and the

962
00:43:59,679 --> 00:44:01,719
departments of health have
not insisted

963
00:44:02,480 --> 00:44:04,239
on the same level of

964
00:44:04,239 --> 00:44:07,199
verification for the surgical
simulators?

965
00:44:08,159 --> 00:44:08,960
Two things.

966
00:44:09,960 --> 00:44:12,199
A surgeon can only kill one
person at a time.

967
00:44:14,079 --> 00:44:18,000
And it's only one person's worth
as opposed to hundreds of

968
00:44:18,000 --> 00:44:19,559
millions of dollars
for a plane

969
00:44:19,880 --> 00:44:22,039
and however many hundreds of people
that are going to sue

970
00:44:22,039 --> 00:44:24,199
the airline to be
really blown

971
00:44:24,199 --> 00:44:24,639
about it.

972
00:44:25,199 --> 00:44:25,519
Yeah.

973
00:44:26,000 --> 00:44:26,199
Yeah.

974
00:44:26,199 --> 00:44:31,599
But I mean, somebody cited in
that PhD, that globally,

975
00:44:31,760 --> 00:44:35,360
there's the number
of deaths due

976
00:44:35,679 --> 00:44:40,800
to surgical misadventure or surgical
skills deficits is

977
00:44:40,800 --> 00:44:43,079
about the equivalent
of one full

978
00:44:43,079 --> 00:44:45,440
jumbo jet going down per day.

979
00:44:45,559 --> 00:44:45,960
Yeah.

980
00:44:46,559 --> 00:44:48,960
And that actually doesn't
surprise me at all.

981
00:44:49,039 --> 00:44:52,159
I mean, since air travel has
gotten so safe these days.

982
00:44:52,559 --> 00:44:55,360
But you know, like I said, even
if it's that many, it's

983
00:44:55,360 --> 00:44:57,639
still one surgeon killing
one person

984
00:44:57,639 --> 00:45:00,159
at a time, widely distributed
as opposed

985
00:45:00,159 --> 00:45:03,679
to 500 people in one
spectacular event.

986
00:45:03,760 --> 00:45:07,599
So unfortunately, it just doesn't
get in people's brains.

987
00:45:07,920 --> 00:45:08,440
Yeah.

988
00:45:08,760 --> 00:45:13,960
I mean, the Bristol case is a
fairly famous case in the

989
00:45:13,960 --> 00:45:16,800
United Kingdom where
there was

990
00:45:16,800 --> 00:45:21,760
a in the Bristol Royal Infirmary,
there were two cardiac,

991
00:45:21,880 --> 00:45:24,440
they were adult cardiac
surgeons

992
00:45:24,440 --> 00:45:27,880
operating on children,
children and their

993
00:45:27,880 --> 00:45:30,559
mortality rate in the Bristol
case report

994
00:45:31,559 --> 00:45:35,960
said that there were there complication,
their mortality

995
00:45:35,960 --> 00:45:37,400
and morbidity was about
twice the

996
00:45:37,400 --> 00:45:38,280
national average.

997
00:45:38,559 --> 00:45:39,960
I'm told privately
that it was

998
00:45:39,960 --> 00:45:41,400
significantly higher
than that.

999
00:45:41,760 --> 00:45:43,920
That went on for the guts
of 10 years.

1000
00:45:44,800 --> 00:45:47,119
And there took 10 years,
right?

1001
00:45:47,119 --> 00:45:47,320
Yeah.

1002
00:45:48,039 --> 00:45:50,360
I mean, their peers knew that
there were issues.

1003
00:45:50,760 --> 00:45:53,119
And it wasn't until it hit
the front pages of Sunday

1004
00:45:53,119 --> 00:45:56,159
newspapers that the
colleges of

1005
00:45:56,159 --> 00:45:57,239
surgeons, the Royal
College of

1006
00:45:57,239 --> 00:45:58,960
Surgeons was forced to
take action.

1007
00:45:59,760 --> 00:46:02,400
You know, that's a
very like I

1008
00:46:02,400 --> 00:46:04,400
said, that was that was
cumulative.

1009
00:46:04,840 --> 00:46:06,360
And it took a long time.

1010
00:46:06,800 --> 00:46:08,679
And I'm sure that there
was a whole lot

1011
00:46:08,679 --> 00:46:11,440
of social stuff going on
at the same time.

1012
00:46:11,480 --> 00:46:11,760
Right.

1013
00:46:11,840 --> 00:46:14,039
Like, yeah, these people
were known

1014
00:46:14,039 --> 00:46:16,039
to other people and blah,
blah, blah.

1015
00:46:16,039 --> 00:46:18,239
And you can't rat
them out or

1016
00:46:18,239 --> 00:46:20,079
whatever, whatever it's
going to be.

1017
00:46:20,079 --> 00:46:22,639
There's all kinds of stuff like,
like I said, back on the

1018
00:46:22,639 --> 00:46:24,400
airplane thing, it's
because it

1019
00:46:24,400 --> 00:46:26,760
all happens at once and the
stakes are so high.

1020
00:46:27,159 --> 00:46:30,480
If it happens one time,
you know, if if a

1021
00:46:30,480 --> 00:46:33,039
surgeon managed to kill
dozens of people in

1022
00:46:33,199 --> 00:46:35,039
the span of a week,
you know, it

1023
00:46:35,039 --> 00:46:36,400
would become a thing,
you know.

1024
00:46:37,239 --> 00:46:37,440
Yeah.

1025
00:46:38,159 --> 00:46:38,400
Yeah.

1026
00:46:38,400 --> 00:46:40,559
There was a wait, wait till
tell it wait till tell us.

1027
00:46:40,599 --> 00:46:43,599
Wait till telesurgery goes on
steroids and you have

1028
00:46:43,599 --> 00:46:46,239
somebody who's doing
way too many

1029
00:46:46,239 --> 00:46:47,039
cases.

1030
00:46:48,119 --> 00:46:50,719
They could make a big mess really
quickly, basically.

1031
00:46:52,079 --> 00:46:52,519
What do you mean?

1032
00:46:53,000 --> 00:46:54,039
Telesurgery goes on steroids.

1033
00:46:54,039 --> 00:46:56,519
You mean that there's a lot
of it being performed or?

1034
00:46:56,920 --> 00:46:57,280
Yeah.

1035
00:46:57,280 --> 00:46:58,519
And you could have a
certain you.

1036
00:46:58,599 --> 00:47:02,199
Well, you and I both know from
being in hospitals that, you

1037
00:47:02,199 --> 00:47:04,079
know, the head surgeons
bouncing

1038
00:47:04,079 --> 00:47:07,760
between operating rooms with the
fellows and the residents

1039
00:47:07,760 --> 00:47:09,239
in there doing doing the most

1040
00:47:09,239 --> 00:47:11,719
of the work, you know, and you
places where they have like

1041
00:47:11,719 --> 00:47:13,320
four, five, six operating
rooms

1042
00:47:13,320 --> 00:47:14,440
going at the same time.

1043
00:47:14,880 --> 00:47:16,760
But the the lead surgeon is

1044
00:47:16,760 --> 00:47:19,159
supposedly the lead surgeon
for all those.

1045
00:47:19,440 --> 00:47:21,400
Imagine what happens when you
have telesurgery.

1046
00:47:22,039 --> 00:47:22,360
Yeah.

1047
00:47:23,920 --> 00:47:28,400
I mean, but in the United States,
OK, that goes on, but

1048
00:47:28,400 --> 00:47:29,920
it's really not supposed
to go

1049
00:47:29,920 --> 00:47:32,719
on the lead surgeon supposed

1050
00:47:32,719 --> 00:47:34,679
to be in the room all
the time.

1051
00:47:35,199 --> 00:47:38,039
And they get away with it because
they sign off on the case

1052
00:47:38,039 --> 00:47:39,519
or they perform the important

1053
00:47:39,519 --> 00:47:40,480
part of the case.

1054
00:47:40,880 --> 00:47:43,800
You know, I mean, they wouldn't
get away with that in the

1055
00:47:43,800 --> 00:47:45,400
UK and Ireland, I
don't think.

1056
00:47:46,400 --> 00:47:46,639
Yeah.

1057
00:47:47,119 --> 00:47:47,440
Yeah.

1058
00:47:47,440 --> 00:47:52,679
I mean, there's no two systematic
reviews and meta analysis

1059
00:47:52,960 --> 00:47:54,559
on proficiency based
progression

1060
00:47:54,559 --> 00:47:57,880
training and comparison to
the training outcomes in

1061
00:47:57,880 --> 00:48:00,840
comparison to conventionally
trained quality

1062
00:48:00,840 --> 00:48:02,760
assured approaches
to training.

1063
00:48:03,039 --> 00:48:04,559
And the results are pretty
similar.

1064
00:48:04,760 --> 00:48:06,639
They basically show that
if you're trained

1065
00:48:06,880 --> 00:48:08,920
in a proficiency based
progression approach,

1066
00:48:09,760 --> 00:48:12,760
you're going to make between
60 and 58 percent fewer

1067
00:48:13,039 --> 00:48:14,559
objectively assessed
interoperative

1068
00:48:14,639 --> 00:48:15,159
errors.

1069
00:48:15,400 --> 00:48:18,920
And there's small scale evidence
that basically said and

1070
00:48:19,159 --> 00:48:20,880
the differences in clinical
outcomes

1071
00:48:20,880 --> 00:48:22,440
is about 60 percent as well.

1072
00:48:23,119 --> 00:48:26,480
So why do you think
that the PBP

1073
00:48:26,480 --> 00:48:31,039
approach hasn't been hasn't
had wider adoption?

1074
00:48:31,760 --> 00:48:36,079
I mean, this is 20 odd years
now, 24 years.

1075
00:48:36,800 --> 00:48:39,039
Well, I would guess, as
you well know,

1076
00:48:39,079 --> 00:48:41,480
it's a lot of work to do a
curriculum based on

1077
00:48:41,480 --> 00:48:41,920
this.

1078
00:48:42,360 --> 00:48:44,000
I mean, it's a whole lot
of work trying

1079
00:48:44,000 --> 00:48:47,559
to figure out what the
metrics are and what

1080
00:48:47,920 --> 00:48:49,880
progression is like in
an objective

1081
00:48:49,880 --> 00:48:52,920
manner to actually be able
to lay that out.

1082
00:48:52,920 --> 00:48:54,360
And it's a lot of work.

1083
00:48:54,679 --> 00:48:57,159
So I think that's a lot
of the problem.

1084
00:48:57,159 --> 00:49:00,400
And I think the other thing is
that I suspect that people

1085
00:49:00,400 --> 00:49:01,679
don't like the idea of having

1086
00:49:01,679 --> 00:49:05,119
to train to some kind of objective
threshold because they

1087
00:49:05,119 --> 00:49:07,079
have no idea how long
it's going

1088
00:49:07,079 --> 00:49:09,119
to take them to attain
it, frankly.

1089
00:49:11,119 --> 00:49:14,599
But the evidence that evidence
is out there now, you can

1090
00:49:14,599 --> 00:49:16,320
train somebody to proficiency

1091
00:49:16,519 --> 00:49:18,079
in one third of the
time than you

1092
00:49:18,079 --> 00:49:20,480
can using a conventional
approach.

1093
00:49:20,840 --> 00:49:25,159
And it's about less than
half of the cost

1094
00:49:25,519 --> 00:49:27,360
to train somebody using
the proficiency.

1095
00:49:27,760 --> 00:49:29,599
And I mean, that data
is published.

1096
00:49:30,320 --> 00:49:30,639
Yeah.

1097
00:49:30,679 --> 00:49:31,039
OK.

1098
00:49:31,639 --> 00:49:34,039
Well, maybe it's just the
words not gotten around.

1099
00:49:34,280 --> 00:49:35,280
This is too new.

1100
00:49:35,639 --> 00:49:36,679
People aren't listening.

1101
00:49:36,719 --> 00:49:38,639
I don't know what to
tell you there.

1102
00:49:40,199 --> 00:49:44,800
But it's well, it's not it's too
new in the general sense.

1103
00:49:44,880 --> 00:49:47,239
I mean, they've been teaching
people the same way for

1104
00:49:48,440 --> 00:49:50,920
hundreds of years,
if you will.

1105
00:49:51,440 --> 00:49:54,679
So this is still kind of
working its way.

1106
00:49:54,679 --> 00:49:58,280
I don't disagree with you because
I'm a big supporter of

1107
00:49:58,400 --> 00:49:59,719
proficiency based
progression, as

1108
00:49:59,719 --> 00:50:00,239
you know.

1109
00:50:01,360 --> 00:50:03,920
I think maybe one of the other
things is that it needs to

1110
00:50:03,920 --> 00:50:06,920
be made more readily
available

1111
00:50:06,920 --> 00:50:09,320
in some kind of an automation
way of doing it.

1112
00:50:09,480 --> 00:50:12,039
You know, you and I have talked
about that, like how you

1113
00:50:12,039 --> 00:50:13,679
could leverage machine
learning

1114
00:50:13,679 --> 00:50:15,239
to actually make it so
that you can

1115
00:50:15,239 --> 00:50:19,880
actually do it without having
to train people to do

1116
00:50:19,880 --> 00:50:21,320
the evaluations and
the likes,

1117
00:50:21,320 --> 00:50:23,119
which is one of the negating
factors.

1118
00:50:23,119 --> 00:50:23,400
Right.

1119
00:50:23,920 --> 00:50:24,320
Yeah.

1120
00:50:24,840 --> 00:50:26,480
Well, I mean, that's
something

1121
00:50:26,480 --> 00:50:29,239
still in the in the planning
process.

1122
00:50:29,519 --> 00:50:33,559
But I mean, I was asked the question
in Australia, I was at

1123
00:50:33,559 --> 00:50:37,880
the surgical robotics
meeting in

1124
00:50:37,880 --> 00:50:40,559
Melbourne and somebody asked
me, Tony, if

1125
00:50:40,559 --> 00:50:41,880
you've got all this, I
was talking about

1126
00:50:41,880 --> 00:50:44,079
proficiency based progression
training and robotics.

1127
00:50:44,440 --> 00:50:45,920
He said, Tony, if you've
got all this

1128
00:50:45,960 --> 00:50:50,960
evidence, why has it not
had wider adoption?

1129
00:50:51,159 --> 00:50:53,400
And I hadn't even thought
about the question.

1130
00:50:53,679 --> 00:50:55,320
But my answer was out before.

1131
00:50:55,639 --> 00:50:57,559
And clearly, I must have had
it in the back of my mind.

1132
00:50:57,559 --> 00:50:59,239
I said, a failure of
leadership.

1133
00:51:00,079 --> 00:51:02,159
I mean, the reason that
Orsi Academy, the

1134
00:51:02,159 --> 00:51:04,199
reason that the Arthroscopic
Association of

1135
00:51:04,199 --> 00:51:08,760
North America and to some
extent, now, Medtronic,

1136
00:51:09,440 --> 00:51:11,159
cardiovascular seems to
be going down

1137
00:51:11,159 --> 00:51:11,639
that road.

1138
00:51:12,079 --> 00:51:14,159
The reason is somebody
senior in the

1139
00:51:14,159 --> 00:51:17,599
organization said, OK, this
is what we're doing.

1140
00:51:18,320 --> 00:51:19,679
And this is not negotiable.

1141
00:51:20,400 --> 00:51:23,960
And I mean, but there doesn't
seem to be enough.

1142
00:51:24,360 --> 00:51:28,360
ERUS, The European robotic
urological society.

1143
00:51:28,760 --> 00:51:34,719
I mean, I showed the data on the
failure rate of the ARIS

1144
00:51:34,800 --> 00:51:35,960
fellowship program,
the number

1145
00:51:35,960 --> 00:51:37,599
of fellows that weren't
finishing.

1146
00:51:37,920 --> 00:51:39,199
And the president, Alberto

1147
00:51:39,199 --> 00:51:42,159
Breda, said, OK, we're
changing.

1148
00:51:42,159 --> 00:51:44,360
We're going to go and use a proficiency
based progression.

1149
00:51:44,880 --> 00:51:46,360
That's what I call
leadership.

1150
00:51:46,639 --> 00:51:49,840
I mean, but it's surgeries,
surgery.

1151
00:51:50,320 --> 00:51:52,000
They must have to have
more leaders.

1152
00:51:52,000 --> 00:51:54,360
And is it something
I'm missing?

1153
00:51:58,079 --> 00:52:00,679
I'm only an engineer as
far as that goes.

1154
00:52:00,679 --> 00:52:03,119
Yeah, but you've been around
this for a long time,

1155
00:52:03,119 --> 00:52:05,599
and I know that you're using
these things.

1156
00:52:06,880 --> 00:52:08,760
Yeah, well, you know,
a lot of people,

1157
00:52:10,599 --> 00:52:14,760
yeah, no, I'm just assuming
that they

1158
00:52:14,960 --> 00:52:17,079
don't think they need it for
whatever reason.

1159
00:52:18,119 --> 00:52:21,800
I mean, you've seen, you know,
in a few isolated cases,

1160
00:52:21,800 --> 00:52:23,119
you've seen people make
the decision that

1161
00:52:23,119 --> 00:52:25,280
they do need it, and that's
that, right?

1162
00:52:25,639 --> 00:52:27,280
I mean, it might also be, I

1163
00:52:27,280 --> 00:52:28,679
mean, like with flight
simulation,

1164
00:52:28,679 --> 00:52:30,719
I mean, that's all government
regulated.

1165
00:52:30,840 --> 00:52:33,800
It's mandated, you know,
and so you'd need

1166
00:52:33,800 --> 00:52:37,199
something like that, but I don't
know if that'll ever

1167
00:52:37,199 --> 00:52:38,199
certainly won't happen
in the United

1168
00:52:38,199 --> 00:52:39,960
States anytime soon, put it

1169
00:52:39,960 --> 00:52:44,039
mildly. But yeah, I would
think it.

1170
00:52:45,559 --> 00:52:49,239
Well, our current environment,
which insists that the

1171
00:52:49,239 --> 00:52:50,719
government can't do anything
right,

1172
00:52:50,719 --> 00:52:53,840
and all regulations are evil
for some reason, you know.

1173
00:52:55,159 --> 00:52:58,000
Yeah, I was in, it was a long

1174
00:52:58,039 --> 00:53:01,239
time ago, 2006, I was
in Australia,

1175
00:53:01,239 --> 00:53:02,440
giving a talk for the Royal

1176
00:53:02,440 --> 00:53:04,159
Australasian College
of Surgeons,

1177
00:53:04,519 --> 00:53:08,559
and part of the deal was that
I would talk to the aviation

1178
00:53:08,559 --> 00:53:10,400
industry about simulation
based training.

1179
00:53:10,719 --> 00:53:16,320
This was 2006, and one
of the leaders

1180
00:53:16,320 --> 00:53:18,679
of the aviation simulation
group come up,

1181
00:53:18,960 --> 00:53:20,679
he says to me, "Hold on
a second." He

1182
00:53:20,679 --> 00:53:22,760
said, "You have all this
validation data

1183
00:53:23,559 --> 00:53:27,000
about how effective it is in terms
of skills training, and

1184
00:53:27,000 --> 00:53:28,440
you can't, it's not adopted?"

1185
00:53:28,679 --> 00:53:32,639
And I said, "No." And he said,
"How come you've got it?" He

1186
00:53:32,639 --> 00:53:34,760
says, "We had no evidence
when it was

1187
00:53:34,760 --> 00:53:37,519
mandated." Somebody
in government

1188
00:53:37,519 --> 00:53:40,239
basically decided, "Okay,
this is a better way."

1189
00:53:41,639 --> 00:53:46,440
And I mean, I don't know how we
persuade them. A fellow of

1190
00:53:46,440 --> 00:53:48,760
mine said to me, "Tony,
this isn't

1191
00:53:49,480 --> 00:53:52,519
data and evidence thing.
This is a matter

1192
00:53:52,519 --> 00:53:55,920
of selling the idea to the
powers that be."

1193
00:53:58,079 --> 00:54:00,880
So you do a lot of work
and publish on

1194
00:54:00,880 --> 00:54:03,480
the developments in robotics
and simulation.

1195
00:54:04,559 --> 00:54:05,679
Where do you envision that

1196
00:54:05,679 --> 00:54:07,719
simulation based training
and robotics

1197
00:54:08,000 --> 00:54:10,360
are heading in the next
10 years or so?

1198
00:54:11,599 --> 00:54:14,519
Well, just like everything else,
it's going to all change

1199
00:54:14,519 --> 00:54:16,800
completely because of AI,
quite honestly.

1200
00:54:18,079 --> 00:54:22,159
I do think you'll see a lot more
enhancements of things,

1201
00:54:22,800 --> 00:54:24,559
like the simulators
because of AI

1202
00:54:24,599 --> 00:54:30,320
as time moves on. It's definitely
going to happen. I think

1203
00:54:30,360 --> 00:54:31,880
it'll become more pervasive
too,

1204
00:54:31,880 --> 00:54:37,480
because the threshold to do it...
So I could go out and

1205
00:54:37,480 --> 00:54:40,159
build a new simulator
now by myself

1206
00:54:40,159 --> 00:54:44,199
because of AI. I'm not going
to do it because it doesn't

1207
00:54:44,199 --> 00:54:46,079
really make a lot of sense
economically,

1208
00:54:46,440 --> 00:54:50,280
but I'd have the ability to do
it because of how well Claude

1209
00:54:50,280 --> 00:54:53,440
and OpenAI Codex and
all these other

1210
00:54:53,440 --> 00:54:56,760
things work now. It's going to
dramatically drop the bar.

1211
00:54:57,199 --> 00:54:59,159
Robots are a different
issue because you

1212
00:54:59,159 --> 00:55:01,199
still got to go through all
the development of the

1213
00:55:01,199 --> 00:55:05,679
mechanics and things. It's
not like AI is going

1214
00:55:05,679 --> 00:55:09,280
to make more surgical robots
pop out quicker, but you're

1215
00:55:09,280 --> 00:55:11,400
going to see a lot more
software enablement

1216
00:55:11,400 --> 00:55:14,480
in surgical robotics.
So there'll be

1217
00:55:14,480 --> 00:55:17,800
little automations, little
part-task automations

1218
00:55:18,440 --> 00:55:21,559
that are going to start showing
up. Same in surgical

1219
00:55:21,559 --> 00:55:24,039
simulation, you'll see
a lot more

1220
00:55:24,079 --> 00:55:27,119
sort of semi-automated instruction
things are going to

1221
00:55:27,119 --> 00:55:28,800
start showing up independent
of the

1222
00:55:28,800 --> 00:55:31,920
simulator cores themselves.
Yeah, you're

1223
00:55:31,920 --> 00:55:33,920
just going to see a lot more
of this for sure.

1224
00:55:33,960 --> 00:55:39,760
And how fast, do you think
that autonomous robotics

1225
00:55:39,920 --> 00:55:42,039
performance surgical procedures
will develop?

1226
00:55:42,039 --> 00:55:49,119
No, no, just no. So you've
heard me say

1227
00:55:49,119 --> 00:55:50,519
this, other people have
heard me say this,

1228
00:55:50,840 --> 00:55:55,159
I'm a real downer on this relative
to academics because

1229
00:55:55,159 --> 00:55:56,880
they're all wanting to
do this, is that

1230
00:55:57,239 --> 00:56:01,159
unless the financial structures
are changed, there's no way

1231
00:56:01,400 --> 00:56:03,559
that autonomous surgery
is going

1232
00:56:03,559 --> 00:56:07,199
to happen with commercial surgical
robots because they're

1233
00:56:07,199 --> 00:56:08,639
not going to allow themselves
to become

1234
00:56:08,639 --> 00:56:13,320
financially liable for what
their robot does. As it is

1235
00:56:13,320 --> 00:56:15,280
right now, they just
build the tool

1236
00:56:15,519 --> 00:56:18,360
and others use the tool and those
other people are the ones

1237
00:56:18,360 --> 00:56:21,239
that become liable from
use of that tool.

1238
00:56:22,239 --> 00:56:26,000
If they start automating things,
then they're doing

1239
00:56:26,039 --> 00:56:28,239
surgery, they're performing
medicine,

1240
00:56:28,320 --> 00:56:32,559
and that's a whole different world.
You'll see lots of AI

1241
00:56:32,559 --> 00:56:34,480
that is going to be
suggestive

1242
00:56:34,880 --> 00:56:37,159
of like, maybe you shouldn't do
that, or maybe you should

1243
00:56:37,159 --> 00:56:38,639
notice this other
thing there,

1244
00:56:40,159 --> 00:56:44,000
stuff like that. You'll see lots
of that, but close the

1245
00:56:44,000 --> 00:56:46,039
wound for me. I don't
think so.

1246
00:56:49,039 --> 00:56:52,920
You might have smart things.
Your stapler example is a

1247
00:56:52,920 --> 00:56:55,440
really good one. There is
a possibility that

1248
00:56:55,559 --> 00:56:59,800
some kind of a supervised stapler
that just performs that

1249
00:57:00,119 --> 00:57:01,679
little tiny task and
does it much

1250
00:57:01,679 --> 00:57:07,679
more uniformly and precisely
than the human does. You're

1251
00:57:07,679 --> 00:57:09,320
not going to see autonomous
procedures.

1252
00:57:09,559 --> 00:57:14,079
You could in a crazy situation
like Satava always used to

1253
00:57:14,119 --> 00:57:15,719
talk about, like battlefield
medicine,

1254
00:57:16,199 --> 00:57:20,159
or you're on your ship to Mars
or whatever it is, then

1255
00:57:20,159 --> 00:57:23,360
totally, I get that. But
in a general sense,

1256
00:57:23,400 --> 00:57:29,400
well, so there might be things
in the developing world that

1257
00:57:29,400 --> 00:57:31,880
this might be a possibility
to,

1258
00:57:31,920 --> 00:57:34,760
but I don't see it in
the Western

1259
00:57:35,159 --> 00:57:43,199
Europe. All the developed
markets, China, Asia,

1260
00:57:43,679 --> 00:57:46,639
Europe, the United States, I
don't see it. I just don't.

1261
00:57:47,519 --> 00:57:50,440
Yeah. I'm sure that the
clinicians will

1262
00:57:50,440 --> 00:57:52,119
have something to say about
it as well.

1263
00:57:52,320 --> 00:57:56,679
I mean, I assume they're particularly
aware of the

1264
00:57:58,719 --> 00:58:00,840
perceived threat of
autonomous

1265
00:58:01,199 --> 00:58:05,559
operators or do they have a realistic
understanding? This

1266
00:58:05,559 --> 00:58:06,480
is probably not going
to happen.

1267
00:58:07,840 --> 00:58:12,400
Yeah. Like I said, I think
it's a regulatory and a

1268
00:58:12,440 --> 00:58:15,159
financial issue and it's
not a technology

1269
00:58:15,159 --> 00:58:17,280
issue. I think they will
figure out how to

1270
00:58:17,280 --> 00:58:22,920
do it, but it's just like
with self-driving,

1271
00:58:22,920 --> 00:58:25,920
right? The self-driving cars,
they get freaked out by

1272
00:58:25,920 --> 00:58:27,599
things they haven't seen
before, right?

1273
00:58:28,159 --> 00:58:30,480
And they've been trying
really hard for a

1274
00:58:30,480 --> 00:58:33,000
really long time to get it
to all be perfect,

1275
00:58:33,000 --> 00:58:37,119
and it's still not perfect. And
you'd have to do the same

1276
00:58:37,119 --> 00:58:39,639
thing in surgery, and you
just don't have

1277
00:58:39,679 --> 00:58:44,119
the infrastructure that you have
for autonomous driving in

1278
00:58:44,119 --> 00:58:45,960
the first place, right?
In terms of

1279
00:58:45,960 --> 00:58:49,280
data collection. And as we know,
there's millions of miles

1280
00:58:49,280 --> 00:58:51,239
of simulated driving that's
used to train

1281
00:58:51,239 --> 00:58:55,159
the autonomous driving systems.
You'd have to do the same

1282
00:58:55,159 --> 00:58:59,920
thing for surgical robotics.
And that

1283
00:58:59,920 --> 00:59:02,679
would mean that you'd have to
verify the simulators being

1284
00:59:02,679 --> 00:59:05,239
accurate so that you know
that the footage

1285
00:59:05,239 --> 00:59:07,280
you're synthetically
generating is

1286
00:59:07,280 --> 00:59:09,480
actually reflective of what
happens in real life.

1287
00:59:09,920 --> 00:59:12,639
So that might be the only way
that we end up with good

1288
00:59:12,639 --> 00:59:14,880
surgical simulators. If
somebody decides

1289
00:59:14,880 --> 00:59:17,800
they're going to invest a huge
amount of money into

1290
00:59:17,840 --> 00:59:21,480
generating millions of hours
of surgical video

1291
00:59:22,079 --> 00:59:25,079
through simulation, they'd have
to actually verify the

1292
00:59:25,079 --> 00:59:27,119
Dickens out of it. So it
might be the only way

1293
00:59:27,119 --> 00:59:29,760
you're ever going to get a properly
verified surgical

1294
00:59:29,760 --> 00:59:32,880
simulator. But okay. So
you would have

1295
00:59:32,880 --> 00:59:37,559
a map and the physical forces
that are generated by the

1296
00:59:37,559 --> 00:59:40,320
different behaviors.
I'm saying the

1297
00:59:40,320 --> 00:59:44,400
endovascular again, because
I mean, you're working in a

1298
00:59:44,400 --> 00:59:46,880
confined space. And
I suppose the

1299
00:59:47,480 --> 00:59:50,800
metrics for that there would
be easier. I'm not saying

1300
00:59:50,840 --> 00:59:52,519
easy, but easier to generate.

1301
00:59:53,519 --> 00:59:56,159
But you would still need
the metrics

1302
00:59:56,159 --> 00:59:58,039
of the performance of
the procedure.

1303
00:59:58,639 --> 01:00:01,679
You know what the cardiologist or the
radiologist does what they need to.

1304
01:00:02,880 --> 01:00:06,119
Yeah, so you'd be we'd be verified.
We'd be building

1305
01:00:06,559 --> 01:00:10,400
physical stand ins for
the for the

1306
01:00:10,679 --> 01:00:12,719
for the patient and then
instrumenting the

1307
01:00:12,719 --> 01:00:15,840
hell out of them to get like
the forces and

1308
01:00:15,840 --> 01:00:18,480
torques and pressures and everything
else. And then we'd be

1309
01:00:18,480 --> 01:00:21,920
using those to verify
the actual,

1310
01:00:21,960 --> 01:00:24,000
you know, virtual simulations,
the other

1311
01:00:24,000 --> 01:00:26,960
computational simulations.
And from there,

1312
01:00:26,960 --> 01:00:30,760
you would take those simulators
used by clinicians, and

1313
01:00:30,760 --> 01:00:33,599
you'd probably be back
mapping the you'd

1314
01:00:33,599 --> 01:00:35,599
have them perform the procedures
people that everybody

1315
01:00:35,599 --> 01:00:38,639
agrees are good. And
then you'd be

1316
01:00:38,639 --> 01:00:41,440
back mapping the information
you pull out of those

1317
01:00:41,440 --> 01:00:44,559
simulators and turning those
into metrics,

1318
01:00:44,599 --> 01:00:49,079
basically. So you don't think
that the clinicians have

1319
01:00:49,079 --> 01:00:51,599
anything to worry about about
an autonomous

1320
01:00:51,639 --> 01:00:55,079
robot anytime soon?
No, I really

1321
01:00:55,079 --> 01:00:59,519
don't. I don't see it.
Yeah. Okay.

1322
01:00:59,519 --> 01:01:03,440
You will see part task automation, but
autonomous robots, no

1323
01:01:03,440 --> 01:01:08,119
Yeah, yeah. I mean, I think
that's entirely feasible.

1324
01:01:08,760 --> 01:01:11,679
OK, Dwight, thanks for
joining me and

1325
01:01:11,719 --> 01:01:12,320
Yeah Sure!

1326
01:01:12,320 --> 01:01:15,079
thanks for a very interesting
conversation.

1327
01:01:15,599 --> 01:01:16,480
Yep, absolutely.