1
00:00:04,538 --> 00:00:07,188
I was wondering
if there has been any talk of

2
00:00:07,188 --> 00:00:07,541
using

3
00:00:07,541 --> 00:00:10,777
AI somehow in IRBs especially

4
00:00:10,777 --> 00:00:13,814
to maybe expedite things, or I
don't.

5
00:00:13,814 --> 00:00:16,984
I don't know if there's anything
I could do, but they're there.

6
00:00:16,984 --> 00:00:18,819
So the short answer is yes.

7
00:00:18,819 --> 00:00:23,969
There's a lot of chatter,
for lack of a better word of use

8
00:00:23,969 --> 00:00:24,324
ING

9
00:00:24,625 --> 00:00:27,327
AI to do a variety of things.

10
00:00:27,327 --> 00:00:30,430
I actually heard the tail
end of a talk the other day.

11
00:00:30,797 --> 00:00:33,967
So you will learn next week

12
00:00:33,967 --> 00:00:36,970
that there is a new requirement

13
00:00:37,571 --> 00:00:39,912
as it relates to the informed
consent

14
00:00:39,912 --> 00:00:42,442
form, that at the very top of
the form,

15
00:00:42,442 --> 00:00:46,656
there is key information
that is the most important

16
00:00:46,656 --> 00:00:47,648
information

17
00:00:47,648 --> 00:00:50,651
that the potential participant
should know.

18
00:00:51,051 --> 00:00:55,594
And so a group at the University
of Michigan has basically

19
00:00:55,594 --> 00:00:56,690
created a way

20
00:00:57,090 --> 00:01:00,227
with queries to chat GPT four or
5

21
00:01:00,227 --> 00:01:03,630
or 6 to say, here's the consent
form,

22
00:01:04,298 --> 00:01:07,568
please pull out the key
information

23
00:01:07,568 --> 00:01:11,405
and put it in a readable,
you know, eighth grade language.

24
00:01:11,838 --> 00:01:15,509
And, they sort of attested
that for a while.

25
00:01:15,509 --> 00:01:17,911
They haven't adopted it as a
service

26
00:01:17,911 --> 00:01:19,446
for all investigators.

27
00:01:19,446 --> 00:01:22,672
But, that's one way that
investigators

28
00:01:22,672 --> 00:01:26,153
are using it to comply with
regulations.

29
00:01:27,120 --> 00:01:28,255
There has also been

30
00:01:28,255 --> 00:01:32,259
some chatter about,
could you use AI to do that

31
00:01:32,259 --> 00:01:34,963
administrative review, for
example,

32
00:01:34,963 --> 00:01:36,663
here is the protocol.

33
00:01:36,663 --> 00:01:40,567
Make sure that it has all these
seven components

34
00:01:40,567 --> 00:01:42,834
and that there's
something that makes sense

35
00:01:42,834 --> 00:01:43,203
below.

36
00:01:43,203 --> 00:01:45,605
So there are a lot of ideas.

37
00:01:45,605 --> 00:01:48,442
There isn't yet any chatter

38
00:01:48,442 --> 00:01:52,379
about replacing the IRB with AI,
but who knows?

39
00:01:54,114 --> 00:01:56,383
Hey, Daniel. Hey,

40
00:01:56,383 --> 00:01:57,751
this is super interesting.

41
00:01:57,751 --> 00:02:00,754
My question is about, you
mentioned that

42
00:02:01,221 --> 00:02:04,925
IRBs, vary
in terms of practice and policy,

43
00:02:05,625 --> 00:02:08,628
historically beholden
to federal regulations.

44
00:02:08,628 --> 00:02:10,831
But I'm curious,

45
00:02:10,831 --> 00:02:14,067
how much do they vary from
institution to institution?

46
00:02:14,067 --> 00:02:16,219
Do they look like, vastly
different

47
00:02:16,219 --> 00:02:18,372
or is it generally pretty
similar?

48
00:02:18,572 --> 00:02:19,606
Great question.

49
00:02:19,606 --> 00:02:23,443
And the the short answer, it's
a bit of a bell curve, right.

50
00:02:23,644 --> 00:02:26,646
So a really good example is

51
00:02:26,913 --> 00:02:29,883
when you're doing pediatric
research,

52
00:02:30,450 --> 00:02:33,887
the IRB must categorize research

53
00:02:33,887 --> 00:02:36,890
as no more than minimal risk,

54
00:02:37,591 --> 00:02:40,327
more than minimal risk with,

55
00:02:40,327 --> 00:02:43,463
a, prospect of direct benefit

56
00:02:43,463 --> 00:02:46,833
and more than minimal risk
and potential benefit

57
00:02:46,833 --> 00:02:50,003
to the group of children
represented by the population.

58
00:02:50,003 --> 00:02:50,670
You're enrolling.

59
00:02:51,605 --> 00:02:54,231
So that definition of minimal
risk

60
00:02:54,231 --> 00:02:55,776
is really important

61
00:02:56,009 --> 00:02:58,984
as it relates to what category
of research

62
00:02:58,984 --> 00:02:59,479
it is.

63
00:03:01,181 --> 00:03:02,682
If there's a lot

64
00:03:02,682 --> 00:03:05,294
if there's a lot of risk, right,
there has to be some direct

65
00:03:05,294 --> 00:03:05,686
benefit.

66
00:03:06,353 --> 00:03:08,618
And in the third category, you
know,

67
00:03:08,618 --> 00:03:10,757
you have to be able to say,
well,

68
00:03:10,757 --> 00:03:14,396
what how much risk is too much
to expose,

69
00:03:14,396 --> 00:03:14,928
etc..

70
00:03:15,562 --> 00:03:18,932
So, Sima Sha, who was a fellow

71
00:03:18,932 --> 00:03:21,119
here at the Department of
Bioethics

72
00:03:21,119 --> 00:03:22,869
many years ago, did a study

73
00:03:22,869 --> 00:03:27,674
where she asked IRBs
for pediatric research.

74
00:03:28,008 --> 00:03:30,710
How do you define minimal risk?

75
00:03:30,710 --> 00:03:34,047
And here's a list of potential
things.

76
00:03:34,247 --> 00:03:36,430
Which would you call minimal
risk

77
00:03:36,430 --> 00:03:37,951
and not so blood draw?

78
00:03:39,152 --> 00:03:42,856
You know, up to like,
a lumbar puncture or,

79
00:03:43,924 --> 00:03:46,159
take an amount of blood, right.

80
00:03:46,159 --> 00:03:47,527
She sort of said so.

81
00:03:47,527 --> 00:03:50,063
Bottom line is
there was a lot of variation,

82
00:03:50,063 --> 00:03:52,499
but it was about a
it was a bell curve, right?

83
00:03:52,499 --> 00:03:56,336
Some IRBs would say
none of that is minimal risk,

84
00:03:56,336 --> 00:04:00,176
even a simple blood draw
where at the other end, lumbar

85
00:04:00,176 --> 00:04:00,874
puncture?

86
00:04:00,874 --> 00:04:02,843
No, not more than.

87
00:04:02,843 --> 00:04:04,511
Again, that's a little extreme,
but

88
00:04:05,612 --> 00:04:06,746
you get the idea.

89
00:04:06,746 --> 00:04:09,583
So it's within a bell curve.

90
00:04:09,583 --> 00:04:13,487
And we as ethics types
might worry about either end.

91
00:04:13,487 --> 00:04:13,887
Right.

92
00:04:13,887 --> 00:04:16,603
So we might worry about exposing
kids

93
00:04:16,603 --> 00:04:17,924
to too much risk,

94
00:04:18,291 --> 00:04:20,829
or we might worry about valuable
research

95
00:04:20,829 --> 00:04:22,129
that isn't happening

96
00:04:22,129 --> 00:04:23,697
because they're so strict.

97
00:04:23,697 --> 00:04:26,433
And that's generally true when
you look

98
00:04:26,433 --> 00:04:31,004
at the literature
about it's not very wide,

99
00:04:31,004 --> 00:04:35,566
but there is some literature
about how IRBs have, you know,

100
00:04:35,566 --> 00:04:35,876
did

101
00:04:35,876 --> 00:04:39,455
some said that is it's exempt,
others said it was expedited,

102
00:04:39,455 --> 00:04:39,813
etc..

103
00:04:39,813 --> 00:04:42,816
So it's almost always a bell
curve,

104
00:04:43,083 --> 00:04:46,052
and very little on either end

105
00:04:46,686 --> 00:04:47,220
and or one.

106
00:04:47,220 --> 00:04:48,922
Go ahead. Hi.

107
00:04:48,922 --> 00:04:51,925
I just have a question
about quality control.

108
00:04:52,259 --> 00:04:55,968
So, I imagine that, like, look,
there are a number of local

109
00:04:55,968 --> 00:04:56,596
libraries

110
00:04:56,596 --> 00:04:58,331
and they all have
a bunch of different people.

111
00:04:58,331 --> 00:04:59,566
Who sit on these committees.

112
00:04:59,566 --> 00:05:02,563
So is there a procedure for
these arabes

113
00:05:02,563 --> 00:05:03,537
to sit like,

114
00:05:03,537 --> 00:05:06,504
step back and look at all
the cases that they've, you

115
00:05:06,504 --> 00:05:06,840
know.

116
00:05:07,741 --> 00:05:09,976
Presided over
and looked at in the past, you.

117
00:05:09,976 --> 00:05:12,812
Know, a number of months
and then, make some sort.

118
00:05:12,812 --> 00:05:14,347
Of determination about whether
their.

119
00:05:14,347 --> 00:05:16,683
Decisions have been generally
consistent.

120
00:05:16,683 --> 00:05:17,450
Yeah.

121
00:05:17,450 --> 00:05:20,554
So this is a topic
that is very close to my heart.

122
00:05:20,554 --> 00:05:23,557
I spend a lot of my time
these days thinking about

123
00:05:23,924 --> 00:05:26,565
how can we evaluate the quality
of IRB

124
00:05:26,565 --> 00:05:28,094
review and oversight.

125
00:05:28,562 --> 00:05:33,600
The short answer is
there are some tools that are

126
00:05:33,600 --> 00:05:34,834
inadequate.

127
00:05:34,834 --> 00:05:38,458
I would say that an IAB could
step back

128
00:05:38,458 --> 00:05:39,573
and review.

129
00:05:39,573 --> 00:05:41,474
They could look at how did we,

130
00:05:42,776 --> 00:05:43,143
you know,

131
00:05:43,143 --> 00:05:46,782
decide let's look at all
our pediatric research and see

132
00:05:46,782 --> 00:05:47,047
how

133
00:05:47,047 --> 00:05:49,533
we decided whether or not we
always said

134
00:05:49,533 --> 00:05:51,585
a blood draw was a minimal risk.

135
00:05:51,585 --> 00:05:53,453
So one could do that.

136
00:05:53,453 --> 00:05:55,887
But IRBs are really busy, don't
have

137
00:05:55,887 --> 00:05:58,658
enough staff, don't have enough
funding.

138
00:05:58,658 --> 00:06:01,403
So it for some, it's a bit of a
luxury

139
00:06:01,403 --> 00:06:03,930
to be able to take that step
back.

140
00:06:04,431 --> 00:06:07,434
But there are

141
00:06:08,068 --> 00:06:12,505
there are a couple of ways
in which institutions improve

142
00:06:12,505 --> 00:06:14,357
the likelihood that the people
around

143
00:06:14,357 --> 00:06:15,809
the table are well situated.

144
00:06:15,809 --> 00:06:19,045
So they're all trained
in a particular way.

145
00:06:19,045 --> 00:06:21,314
They have in-service training.

146
00:06:21,314 --> 00:06:23,531
I'm sure that if a chair had a
concern,

147
00:06:23,531 --> 00:06:24,951
like you were on the IRB

148
00:06:24,951 --> 00:06:26,774
and you fell asleep and you were
never

149
00:06:26,774 --> 00:06:28,021
bringing your reviews in,

150
00:06:28,788 --> 00:06:30,914
they'd probably have a chat with
you and

151
00:06:30,914 --> 00:06:33,093
ask you like the at the
extremes, right?

152
00:06:33,093 --> 00:06:35,391
There's ways that the sort of
local

153
00:06:35,391 --> 00:06:37,163
quality can be maintained.

154
00:06:37,163 --> 00:06:39,849
And but there are ways that you
could step

155
00:06:39,849 --> 00:06:40,233
back.

156
00:06:40,934 --> 00:06:44,671
Some institutions do ad hoc
audits.

157
00:06:45,138 --> 00:06:46,306
Not for cause.

158
00:06:46,306 --> 00:06:46,606
Right.

159
00:06:46,606 --> 00:06:49,309
We're not worried
about a particular person's

160
00:06:49,309 --> 00:06:49,909
protocol.

161
00:06:50,143 --> 00:06:51,611
You can't review everything.

162
00:06:51,611 --> 00:06:54,263
So you say, I'm going to
randomly select

163
00:06:54,263 --> 00:06:56,583
ten projects and take a deep
dive.

164
00:06:56,850 --> 00:06:59,331
And so those audits can then
inform

165
00:06:59,331 --> 00:07:01,955
wow, everybody's forgetting to
do X.

166
00:07:01,955 --> 00:07:04,381
So we've got to reorient the
investigators

167
00:07:04,381 --> 00:07:05,825
and the IRB to make sure

168
00:07:05,825 --> 00:07:06,626
this is present.

169
00:07:08,828 --> 00:07:09,629
Go ahead.

170
00:07:09,629 --> 00:07:12,632
Hi, Holly. Sally Fowler. Hi
there.

171
00:07:12,632 --> 00:07:14,691
I kind of more practical
question

172
00:07:14,691 --> 00:07:15,502
on something

173
00:07:15,502 --> 00:07:17,289
you just put up, and it's about
irb

174
00:07:17,289 --> 00:07:18,872
used in a responsible parties.

175
00:07:18,872 --> 00:07:23,273
And you wrote that, IRBs
have to do the continuing

176
00:07:23,273 --> 00:07:23,977
review.

177
00:07:23,977 --> 00:07:27,128
Yes. On studies
that receive full committee

178
00:07:27,128 --> 00:07:28,081
review. Yep.

179
00:07:28,081 --> 00:07:31,648
So expedited reviews
which, you know, are minimal

180
00:07:31,648 --> 00:07:32,085
risk.

181
00:07:32,085 --> 00:07:33,653
Yeah.

182
00:07:33,653 --> 00:07:36,050
Are you saying
because this would be kind of

183
00:07:36,050 --> 00:07:36,323
good

184
00:07:36,323 --> 00:07:38,537
if you are
that they don't need continuing

185
00:07:38,537 --> 00:07:38,958
review.

186
00:07:38,958 --> 00:07:41,306
And I know your your caveats
going to be

187
00:07:41,306 --> 00:07:42,362
go to your local.

188
00:07:42,362 --> 00:07:43,563
Yeah I heard that.

189
00:07:43,563 --> 00:07:44,864
So it's a great question.

190
00:07:44,864 --> 00:07:47,727
So one thing none of us
mentioned today

191
00:07:47,727 --> 00:07:50,370
is that the common rule was
revised

192
00:07:50,370 --> 00:07:54,074
in 2018
after a very long, protracted

193
00:07:54,074 --> 00:07:54,974
process.

194
00:07:55,442 --> 00:07:58,615
One of the things that's
different

195
00:07:58,615 --> 00:08:01,881
from before 2018 is that only
full

196
00:08:01,881 --> 00:08:05,262
committee reviewed protocols are
required

197
00:08:05,262 --> 00:08:07,654
to provide an annual review.

198
00:08:08,188 --> 00:08:11,391
So if it's expedited, you don't
have to.

199
00:08:11,391 --> 00:08:14,761
Now, some IRBs could say

200
00:08:15,028 --> 00:08:18,198
we want to do more than what
the regulations require,

201
00:08:18,198 --> 00:08:21,768
and we want to review
all expedited research,

202
00:08:21,768 --> 00:08:25,069
but they're no longer required
by federal regulations to do

203
00:08:25,069 --> 00:08:25,405
that.

204
00:08:25,605 --> 00:08:27,829
So when they approve whichever
IRB

205
00:08:27,829 --> 00:08:30,510
you're in, when they approve
your study,

206
00:08:31,478 --> 00:08:32,245
they will tell.

207
00:08:32,245 --> 00:08:33,313
You exactly they.

208
00:08:33,313 --> 00:08:34,948
Want a continuing review every
year.

209
00:08:34,948 --> 00:08:37,183
We're going to want to know
exactly.

210
00:08:37,183 --> 00:08:40,320
Exactly. And it's not required.

211
00:08:40,320 --> 00:08:44,090
And some IRBs will,
do an even shorter period.

212
00:08:44,090 --> 00:08:46,059
So here at the Clinical Center,

213
00:08:46,059 --> 00:08:47,861
I think I'm going to mention
this

214
00:08:47,861 --> 00:08:49,662
in another, in another, lecture.

215
00:08:50,096 --> 00:08:53,733
The vast majority of research
that happens here is phase

216
00:08:53,733 --> 00:08:56,668
one research
very early, first in human

217
00:08:56,668 --> 00:08:57,270
trials.

218
00:08:57,704 --> 00:09:00,030
And when that type of trial is
going on,

219
00:09:00,030 --> 00:09:02,008
the IRB might say, you know
what,

220
00:09:02,208 --> 00:09:04,210
we're a little worried about
that

221
00:09:04,210 --> 00:09:06,212
or we really want to have close,

222
00:09:06,780 --> 00:09:09,176
we want to be very careful about
the risks

223
00:09:09,176 --> 00:09:10,717
that people are exposed to

224
00:09:10,717 --> 00:09:13,349
and want to make sure
that everything is moving

225
00:09:13,349 --> 00:09:13,853
forward.

226
00:09:13,853 --> 00:09:16,322
We want to review you in six
months.

227
00:09:16,322 --> 00:09:19,142
So the IRB always has the option
to ask

228
00:09:19,142 --> 00:09:21,961
for something earlier if they
want to.