AI adoption in hospitals - Michael Page (Unity Health)
In this episode I discuss the ins and outs of AI adoption in hospitals with Michael. Michael is the director of AI commercialization at Unity Health.
Transcript
1
00:00:00,000 --> 00:00:13,400
Hi Michael, it's great to have you with us today.
2
00:00:13,400 --> 00:00:14,400
Yeah, thanks for having me.
3
00:00:14,400 --> 00:00:16,040
It's wonderful to be here.
4
00:00:16,040 --> 00:00:21,280
If you could tell us a bit about your childhood and go as in-depth as you'd like to about
5
00:00:21,280 --> 00:00:23,480
the past to where you are now.
6
00:00:23,480 --> 00:00:25,480
Yeah.
7
00:00:25,480 --> 00:00:27,840
Where do you begin with childhood?
8
00:00:27,840 --> 00:00:34,960
I think maybe, you know, I tell some people this and maybe it's surprising, maybe it's
9
00:00:34,960 --> 00:00:39,120
not because maybe I like I present a certain way or maybe there's some assumptions of
10
00:00:39,120 --> 00:00:41,320
who I am.
11
00:00:41,320 --> 00:00:52,000
Grew up in sort of this like suburban countryside and sort of a lower income working class family.
12
00:00:52,000 --> 00:00:58,920
So I'd say childhood was really, really great filled with creativity and a lot of time outdoors
13
00:00:58,920 --> 00:01:04,760
and not a lot of time in technology, which I think some people sort of assume.
14
00:01:04,760 --> 00:01:09,040
I might be the kid that's like taking apart radios and try to figure out how to put them
15
00:01:09,040 --> 00:01:15,280
back together the same as like my bike or something else around the garage.
16
00:01:15,280 --> 00:01:22,480
So probably the most impressionable thing is being the first person in my immediate family
17
00:01:22,480 --> 00:01:25,240
to go to university.
18
00:01:25,240 --> 00:01:28,440
And so, you know, education was really important.
19
00:01:28,440 --> 00:01:34,640
It was hard to navigate that as being, you know, kind of that outsider.
20
00:01:34,640 --> 00:01:38,840
My parents kind of, you know, steering me as much as they could, but more things being
21
00:01:38,840 --> 00:01:42,160
unknown than known for them.
22
00:01:42,160 --> 00:01:47,040
And you know, that took me to an incredible path of, you know, everyone says like follow
23
00:01:47,040 --> 00:01:48,040
your passion.
24
00:01:48,040 --> 00:01:52,920
So I started my career in the arts and maybe we could talk a little bit about that.
25
00:01:52,920 --> 00:01:57,840
But where I am today is I ended up going back and doing a master's.
26
00:01:57,840 --> 00:02:02,360
I now work in technology and AI of all things.
27
00:02:02,360 --> 00:02:08,400
And for over a year now, I've been at the Ivy Business School at Western University
28
00:02:08,400 --> 00:02:09,400
now teaching.
29
00:02:09,400 --> 00:02:18,440
So my kind of like full circle moment is being this loved and supported working class kid
30
00:02:18,440 --> 00:02:24,560
to be the first person to go to university to now teach in a university, which I think
31
00:02:24,560 --> 00:02:30,080
is just, you know, if not through education, then it has to be Canada's the only place
32
00:02:30,080 --> 00:02:34,040
where you can make that happen is what I believe.
33
00:02:34,040 --> 00:02:35,040
That's amazing to hear.
34
00:02:35,040 --> 00:02:40,360
So let's go back to your time in high school.
35
00:02:40,360 --> 00:02:47,400
And how did you decide to do a bachelor's in arts and talk to me about your experience
36
00:02:47,400 --> 00:02:52,880
there and then your transition to working for the Toronto Symphony?
37
00:02:52,880 --> 00:02:53,880
Yeah.
38
00:02:53,880 --> 00:02:56,400
Great question.
39
00:02:56,400 --> 00:03:05,000
So I end up becoming sort of this like, you know, kind of now comfortable in crowds and
40
00:03:05,000 --> 00:03:08,920
can sort of speak my mind freely.
41
00:03:08,920 --> 00:03:12,400
But that's not where I started in high school.
42
00:03:12,400 --> 00:03:17,400
I think it was for a good period of time, you know, by sweating through my shirt afraid
43
00:03:17,400 --> 00:03:21,360
of my own shadow a little bit too much.
44
00:03:21,360 --> 00:03:28,120
But at some point in high school, I became kind of the theater kid.
45
00:03:28,120 --> 00:03:33,160
And through, you know, those classes, those opportunities, we had like an improv group
46
00:03:33,160 --> 00:03:35,200
and doing some plays.
47
00:03:35,200 --> 00:03:41,840
I kind of found my voice and being able to talk to people and relate to people and maybe
48
00:03:41,840 --> 00:03:48,440
also being more confident being another character rather than being myself is probably part of
49
00:03:48,440 --> 00:03:50,000
the story.
50
00:03:50,000 --> 00:03:56,440
And so when I was choosing universities, what's interesting is I applied in sort of like a
51
00:03:56,440 --> 00:03:58,360
really eclectic way.
52
00:03:58,360 --> 00:04:04,280
So on one hand, there was like fine arts and sort of bachelor arts programs.
53
00:04:04,280 --> 00:04:09,360
And then on the other hand, were business and technology programs.
54
00:04:09,360 --> 00:04:16,200
And I had actually committed accepted to a different university to do a business and
55
00:04:16,200 --> 00:04:17,200
technology.
56
00:04:17,200 --> 00:04:20,800
I think it was a B.com, but I don't remember.
57
00:04:20,800 --> 00:04:26,840
And on the last day when you can decide or would have you back then, well before email,
58
00:04:26,840 --> 00:04:35,640
I got the acceptance letter from the University of Toronto to do a humanities degree.
59
00:04:35,640 --> 00:04:41,360
And on the very last day, my girlfriend at the time was at the university and I decided
60
00:04:41,360 --> 00:04:48,240
to go to U of T and just sort of pursue the arts and pursue what I love.
61
00:04:48,240 --> 00:04:51,280
I am only able to say this in hindsight.
62
00:04:51,280 --> 00:04:54,720
I don't think I appreciate it at the time.
63
00:04:54,720 --> 00:04:56,760
And there's some brilliant people.
64
00:04:56,760 --> 00:05:02,760
Like I think this quote is loosely based on something that Hal Jackman has said, where
65
00:05:02,760 --> 00:05:07,760
humanities degree or the arts teaches you how to think critically.
66
00:05:07,760 --> 00:05:12,320
It's actually not really about analysis of art in and of itself.
67
00:05:12,320 --> 00:05:15,360
It's about how are you structuring thought?
68
00:05:15,360 --> 00:05:21,120
How can you understand what other people believe, think and value?
69
00:05:21,120 --> 00:05:22,920
And then how do you communicate that?
70
00:05:22,920 --> 00:05:27,080
So that was kind of my takeaway.
71
00:05:27,080 --> 00:05:30,720
And I found myself there by circumstance.
72
00:05:30,720 --> 00:05:37,320
I wish I was probably more intentional at the time, but I did walk away with some skills
73
00:05:37,320 --> 00:05:43,880
and some purpose and some this growth mindset of what I wanted to do in the world.
74
00:05:43,880 --> 00:05:49,400
And what prompted the transition to business and then to AI?
75
00:05:49,400 --> 00:05:51,000
That's a good point.
76
00:05:51,000 --> 00:05:57,360
So I started off at a few different arts organizations as you've noted.
77
00:05:57,360 --> 00:06:04,640
So I worked at a theater company in Branson, Ontario that was just opening up called the
78
00:06:04,640 --> 00:06:06,080
Rose Theater.
79
00:06:06,080 --> 00:06:11,600
I was there for the opening was like one of my first internships, bounced around at like,
80
00:06:11,600 --> 00:06:16,600
you know, a children's theater company, an arts service organization for playwrights.
81
00:06:16,600 --> 00:06:23,400
Found myself at the Toronto Symphony, where the symphony felt like the most professional
82
00:06:23,400 --> 00:06:25,360
place I had ever worked at that point.
83
00:06:25,360 --> 00:06:28,120
I'm like very early 20s.
84
00:06:28,120 --> 00:06:33,440
You know, they've got a board, a CEO, there's hundreds of people, there's lots of revenue
85
00:06:33,440 --> 00:06:34,720
flowing in.
86
00:06:34,720 --> 00:06:40,920
And this is the moment where I start to feel and understand and really experience all the
87
00:06:40,920 --> 00:06:44,840
different, you know, common nodes of business.
88
00:06:44,840 --> 00:06:52,240
So marketing and sales and HR and in our case, fundraising.
89
00:06:52,240 --> 00:06:57,560
And how do they function and how does sort of this corporate structure function?
90
00:06:57,560 --> 00:07:04,440
And, you know, I'll say I was good enough at fundraising and an open an opportunity to
91
00:07:04,440 --> 00:07:11,080
return to U of T where I would work as an administrator in their advancement office
92
00:07:11,080 --> 00:07:15,960
and and I'm spending a decade there grew a lot.
93
00:07:15,960 --> 00:07:20,280
I say it's like, you know, I feel like U of T's played a pivotal role where it's like,
94
00:07:20,280 --> 00:07:24,800
you know, your first university where you're becoming an adult and then kind of where like
95
00:07:24,800 --> 00:07:29,000
I actually grew up by by working there for a decade.
96
00:07:29,000 --> 00:07:38,440
But the transition from all of that into business and technology was very much related to, you
97
00:07:38,440 --> 00:07:42,160
know, first my like my management mindset at the time.
98
00:07:42,160 --> 00:07:48,320
And then the second thing, wow, what an incredible opportunity being at U of T talking about AI
99
00:07:48,320 --> 00:07:51,960
and you know, in the 2010s.
100
00:07:51,960 --> 00:07:57,640
So the first thing is my management mindset before I went back and did my MBA at Ivy Business
101
00:07:57,640 --> 00:08:02,800
School was I always kind of kept the mental note of bad managers and the things that they
102
00:08:02,800 --> 00:08:05,400
would do and we're kind of mentally commit to.
103
00:08:05,400 --> 00:08:09,080
I don't want to do this when I manage.
104
00:08:09,080 --> 00:08:12,520
And I kind of got to a point of like, you know, when you're leading fairly large teams
105
00:08:12,520 --> 00:08:18,600
or you know, really big revenue goals or projects or what have you where I didn't really have
106
00:08:18,600 --> 00:08:20,640
a great skill set, right?
107
00:08:20,640 --> 00:08:25,720
Like this skill set based on I should do things, you know, I should not do the things that
108
00:08:25,720 --> 00:08:30,120
I did not like is not really a strong management philosophy.
109
00:08:30,120 --> 00:08:34,800
And so I felt like, you know, this is a huge part of my career.
110
00:08:34,800 --> 00:08:35,880
I'm not done.
111
00:08:35,880 --> 00:08:42,520
I want to have 10x the impact that I've had thus far.
112
00:08:42,520 --> 00:08:48,000
And if there's if there's one thing I need to learn, I need to learn how to lead.
113
00:08:48,000 --> 00:08:56,360
So that's what took me to Ivy Business School and doing my executive MBA there and kind
114
00:08:56,360 --> 00:08:58,440
of learning and growing.
115
00:08:58,440 --> 00:09:03,560
The transition to technology happened at the same time and it had a lot to do with, you
116
00:09:03,560 --> 00:09:07,360
know, the people and the projects and things I was working at U of T.
117
00:09:07,360 --> 00:09:09,080
I'll just name job for a second.
118
00:09:09,080 --> 00:09:14,040
None of these individuals will remember who I am, but having the opportunity to get assigned
119
00:09:14,040 --> 00:09:20,200
to support or be on projects with some, you know, incredible researchers and then now like
120
00:09:20,200 --> 00:09:30,880
founders like Dr. Raquel Erdison, Dr. Jillian Hadfield and a few others where very early
121
00:09:30,880 --> 00:09:38,360
days like in the 2010s, I'm learning about autonomous vehicles and learning about that,
122
00:09:38,360 --> 00:09:40,600
you know, how did they actually learn?
123
00:09:40,600 --> 00:09:41,760
Why are they learning?
124
00:09:41,760 --> 00:09:46,960
Where is this going and kind of conceptualizing the future or that these systems could be really,
125
00:09:46,960 --> 00:09:55,560
really unsafe if they're not managed or we don't think about the ethics and equity is
126
00:09:55,560 --> 00:10:02,720
really what Dr. Hadfield has contributed to AI and our ecosystem.
127
00:10:02,720 --> 00:10:07,040
And so this planted the seed in me where it's like, you know, we're seeing tech everywhere.
128
00:10:07,040 --> 00:10:13,360
I feel like it's the largest growth industry in the world, but being at U of T in that
129
00:10:13,360 --> 00:10:20,840
moment around, you know, these descendants of Dr. Jeffrey Hinton and where U of T has
130
00:10:20,840 --> 00:10:23,520
kind of led the world.
131
00:10:23,520 --> 00:10:27,160
I had in my gut, but I was probably, you know, just surrounded by some incredible people
132
00:10:27,160 --> 00:10:31,200
where it's like AI is going to be really, really big.
133
00:10:31,200 --> 00:10:36,160
So the transition to business is, you know, MBA, this in depth kind of understanding of
134
00:10:36,160 --> 00:10:42,640
seeing research really at the bleeding edge at U of T and then saying, we got to do more
135
00:10:42,640 --> 00:10:45,040
than just publish papers.
136
00:10:45,040 --> 00:10:50,960
We don't need more citations collecting dust on a bookshelf somewhere.
137
00:10:50,960 --> 00:10:56,200
We have a chance as Canada to really own the podium here and let's figure out how we get
138
00:10:56,200 --> 00:11:05,280
this research out of our really amazing research institutions into companies into everyday life
139
00:11:05,280 --> 00:11:07,840
so that we can, you know, grow economically.
140
00:11:07,840 --> 00:11:14,080
We can create more jobs, but that we can contribute to the world in a meaningful way.
141
00:11:14,080 --> 00:11:20,960
So I joined the Vector Institute and that's been, you know, a wonderful organization that
142
00:11:20,960 --> 00:11:28,240
really grew and groomed me and taught me so much about machine learning.
143
00:11:28,240 --> 00:11:33,920
And now I find myself in healthcare where, you know, we're seeing this every single day
144
00:11:33,920 --> 00:11:37,960
in the way that our clinicians are treating patients.
145
00:11:37,960 --> 00:11:45,920
So along eclectic circus of a journey, but I do relate this a little bit back to, you
146
00:11:45,920 --> 00:11:52,360
think to that 17 or 18 year old humanities kid, a lot of this is based there of how can
147
00:11:52,360 --> 00:11:58,560
I understand what other people value or what they communicate and how can I think critically
148
00:11:58,560 --> 00:11:59,640
about these things.
149
00:11:59,640 --> 00:12:06,680
And AI for me is really about that of we have this massive corpus of information.
150
00:12:06,680 --> 00:12:07,680
How do you take it in?
151
00:12:07,680 --> 00:12:09,000
How do you think critically?
152
00:12:09,000 --> 00:12:14,360
How do you exercise judgment or build systems that are effective or safe?
153
00:12:14,360 --> 00:12:18,600
And so the eye rely on the business skills and, you know, and some of the hustle and
154
00:12:18,600 --> 00:12:21,160
acumen and stuff like that that I've built up.
155
00:12:21,160 --> 00:12:26,320
But a lot of this is that arts background of how do we think critically?
156
00:12:26,320 --> 00:12:30,600
How do we understand what the current context is?
157
00:12:30,600 --> 00:12:35,920
Does the common startup philosophy that says that you should hire slow and fire fast, do
158
00:12:35,920 --> 00:12:37,160
you agree with that?
159
00:12:37,160 --> 00:12:41,080
Oh, that's a great question.
160
00:12:41,080 --> 00:12:42,520
I like the higher slow.
161
00:12:42,520 --> 00:12:47,880
So like I think, you know, we're coming off a period where venture back companies got
162
00:12:47,880 --> 00:12:49,640
incredible valuations.
163
00:12:49,640 --> 00:12:54,720
They hired hundreds, sometimes thousands of people and really did not have the runway
164
00:12:54,720 --> 00:13:00,920
to sustainably support that head count.
165
00:13:00,920 --> 00:13:06,360
The fire fasting, I'd say like, you know, it depends on sort of the context.
166
00:13:06,360 --> 00:13:15,080
I don't know what the definition would be around that, but I find that most individuals
167
00:13:15,080 --> 00:13:16,680
are really learning a job.
168
00:13:16,680 --> 00:13:22,080
You're contributing, but you're learning a job probably for the first year.
169
00:13:22,080 --> 00:13:27,600
The second year is where, you know, you're moving beyond sort of this binary learning
170
00:13:27,600 --> 00:13:32,600
like, oh, I was giving a task, I executed the task.
171
00:13:32,600 --> 00:13:37,640
The second year is when you really understand that context, your internal context, and you
172
00:13:37,640 --> 00:13:42,880
understand the external context where you might be able to be a top contributor or not.
173
00:13:42,880 --> 00:13:48,040
Now, all it's that people don't show up, they're late or quality of work, all that, you know,
174
00:13:48,040 --> 00:13:55,280
aside, but I do take a longer term view, depending if that is a long term view for some people,
175
00:13:55,280 --> 00:14:00,720
on human capital in particular, because it organizations are weird.
176
00:14:00,720 --> 00:14:04,040
Everyone is unique and everyone says that they're unique.
177
00:14:04,040 --> 00:14:10,200
And I don't know that enough leaders reflect on how poor their internal contacts might be
178
00:14:10,200 --> 00:14:13,840
for individuals to succeed.
179
00:14:13,840 --> 00:14:17,560
And so I feel like it's around the year mark, sometimes it's shorter, sometimes it's longer,
180
00:14:17,560 --> 00:14:22,360
but I find that sort of that second year is when people really kind of hit their stride
181
00:14:22,360 --> 00:14:25,600
and you can really see the value that they bring.
182
00:14:25,600 --> 00:14:27,480
Have you ever made a bad hire?
183
00:14:27,480 --> 00:14:31,200
And is there something you could have done in the hiring process that would have stopped
184
00:14:31,200 --> 00:14:32,360
you from hiring them?
185
00:14:32,360 --> 00:14:39,520
Yeah, made a bad hire, for sure, multiple times, did not see it coming.
186
00:14:39,520 --> 00:14:45,920
So I think, you know, interviews are getting to be a little bit more interesting and better,
187
00:14:45,920 --> 00:14:54,120
but something that I really, really reflect on, this does, you know, follow the murder
188
00:14:54,120 --> 00:15:01,080
of George Floyd and Candace efforts and truth and reconciliation is, I do think that as leaders,
189
00:15:01,080 --> 00:15:06,720
managers, people hiring, we need to think about how fraught the hiring process is.
190
00:15:06,720 --> 00:15:12,040
And this has nothing to do with AI suggesting like who we should hire, what resumes we should
191
00:15:12,040 --> 00:15:20,360
read, the process has been fraught to hire the person that went to the best school, had
192
00:15:20,360 --> 00:15:25,800
the best network, had the best reference, it's sort of been coached to do the same.
193
00:15:25,800 --> 00:15:33,960
And I'll say like, you know, we've taken it in quote, a chance on a few people where the
194
00:15:33,960 --> 00:15:38,080
interview was clunky, or you know, there was something where we're like, oh, we're not
195
00:15:38,080 --> 00:15:46,760
too sure, you can't see hustle, you can't see commitment, you can't see drive, and
196
00:15:46,760 --> 00:15:51,840
you're 10 standardized questions, and the one or two assignments you give someone, or
197
00:15:51,840 --> 00:15:55,160
the, you know, one to three references that you check.
198
00:15:55,160 --> 00:16:03,800
So who a person is, you know, you might get that sometimes in an interview process.
199
00:16:03,800 --> 00:16:08,640
And that's something where you're building a relationship together, and that spark may
200
00:16:08,640 --> 00:16:09,640
or may not happen.
201
00:16:09,640 --> 00:16:16,160
But if you find someone that really is aligned with your personal purpose or your corporate
202
00:16:16,160 --> 00:16:22,160
purpose, and they will literally move mountains for you, that's a great hire.
203
00:16:22,160 --> 00:16:30,040
And you know, if they're a bit quirky or late or they've got too many cats, who cares?
204
00:16:30,040 --> 00:16:34,760
As long as they're kind of, you know, really devoted to where both of you can go together.
205
00:16:34,760 --> 00:16:38,400
Yeah, my kids would say you can never have too many cats.
206
00:16:38,400 --> 00:16:43,880
I know my son desperately wants cats and dogs.
207
00:16:43,880 --> 00:16:47,680
We had a dog, and we're now in the like in between phase.
208
00:16:47,680 --> 00:16:52,640
But yes, if we had 16 cats, he would love it.
209
00:16:52,640 --> 00:16:56,120
I think people underscore the importance of references.
210
00:16:56,120 --> 00:17:02,400
And there's a couple of ways to do it, I usually seek references from previous jobs that the
211
00:17:02,400 --> 00:17:07,000
founders don't provide readily, and I let them know that.
212
00:17:07,000 --> 00:17:11,240
Or you can ask for 10 references and say you will randomly call three of them.
213
00:17:11,240 --> 00:17:14,000
Now that's a big ask.
214
00:17:14,000 --> 00:17:18,840
And the job offer or the investment has to reflect that.
215
00:17:18,840 --> 00:17:24,680
But you want references where the founder has had conflict as well.
216
00:17:24,680 --> 00:17:30,160
I think so, and you know, and understanding how people navigate conflict, like sometimes
217
00:17:30,160 --> 00:17:32,560
it really shuts someone down.
218
00:17:32,560 --> 00:17:38,400
And other times, someone can take it on the chin and say, you know, thank you.
219
00:17:38,400 --> 00:17:39,400
I disagree.
220
00:17:39,400 --> 00:17:43,040
But I'm going to think about this or I'm going to learn from this, this opportunity.
221
00:17:43,040 --> 00:17:49,840
So I like how you how you navigate that, I think is, is, is quite important.
222
00:17:49,840 --> 00:17:56,560
I think like, you know, I've seen things where like some companies are doing like mixers or
223
00:17:56,560 --> 00:18:03,360
actually the hiring team is kind of not involved for a certain, you know, round or phase.
224
00:18:03,360 --> 00:18:10,120
It's someone on the team is like messages as a person, hey, are you free for coffee?
225
00:18:10,120 --> 00:18:11,440
That's what I'm interested in.
226
00:18:11,440 --> 00:18:16,400
You know, if this person for an hour or two hours or as coach, the references can put
227
00:18:16,400 --> 00:18:20,200
something together for me, that's fine.
228
00:18:20,200 --> 00:18:28,840
But you know, Karen and Kayla, who you're going to have to work with every single day.
229
00:18:28,840 --> 00:18:29,840
Can you get along?
230
00:18:29,840 --> 00:18:35,520
Like is there some sort of animosity or friction or is there actually, you know, some, some
231
00:18:35,520 --> 00:18:38,160
real bond here for, for whatever reason?
232
00:18:38,160 --> 00:18:42,960
Is there sort of that that spark of opportunity?
233
00:18:42,960 --> 00:18:48,320
What is hard, and I do feel like there's some people talking about this, but what is really,
234
00:18:48,320 --> 00:18:59,480
really hard in AI and in tech is we just have many, many engineers and computer scientists
235
00:18:59,480 --> 00:19:01,400
that are introverted.
236
00:19:01,400 --> 00:19:09,360
And so if you look at the traditional scorecard of, you know, character or leadership, education
237
00:19:09,360 --> 00:19:13,880
experience, whatever, like how we're going to score someone, that individual just might
238
00:19:13,880 --> 00:19:20,160
be really, really nervous to talk to you or to share in that way.
239
00:19:20,160 --> 00:19:24,440
So part of like, you know, do we need to rethink this as some of this just like, would you
240
00:19:24,440 --> 00:19:28,440
be open like we just text over the next like five days of like, you know, I'd love to hear
241
00:19:28,440 --> 00:19:34,440
your thoughts on this big announcement from, from the video, like what do you think the
242
00:19:34,440 --> 00:19:42,280
impact would be on, on our work or health Canada just issued updated regulation on AI's medical
243
00:19:42,280 --> 00:19:46,440
device like what do you thought we can ask that and you know, maybe I've been coached
244
00:19:46,440 --> 00:19:51,080
and trained and have more experience where off the cuff, I'm going to give you something
245
00:19:51,080 --> 00:19:56,880
that's that's eloquent, but maybe for that interview to introverted engineer or computer
246
00:19:56,880 --> 00:20:01,400
scientists, let's just do that over email or do that over Slack or do that over text
247
00:20:01,400 --> 00:20:03,160
messages or something else.
248
00:20:03,160 --> 00:20:07,040
And maybe that's the way that we're going to build trust in that relationship.
249
00:20:07,040 --> 00:20:09,240
Maybe that's the way that I should be evaluating you.
250
00:20:09,240 --> 00:20:15,400
So that's a long way of me sort of saying that like, the world has changed, you know,
251
00:20:15,400 --> 00:20:22,920
like we're working remotely, very, very often, have our hiring practices changed as as radically
252
00:20:22,920 --> 00:20:24,200
probably not.
253
00:20:24,200 --> 00:20:28,040
And we probably need to rethink it.
254
00:20:28,040 --> 00:20:29,800
What are your thoughts in remote work?
255
00:20:29,800 --> 00:20:37,920
And you have a similar culture remotely, because a lot of us, a lot of companies are
256
00:20:37,920 --> 00:20:41,720
mandating back in office.
257
00:20:41,720 --> 00:20:44,680
Where do you fall on that spectrum?
258
00:20:44,680 --> 00:20:50,640
So it's like a working parent hybrid is an incredible benefit.
259
00:20:50,640 --> 00:20:55,640
And I can't say that it's that it's not you mentioned you have kids.
260
00:20:55,640 --> 00:20:56,640
They're sick.
261
00:20:56,640 --> 00:20:59,840
All these crazy things that happen with with kids schedules.
262
00:20:59,840 --> 00:21:05,080
I know a lot of us also have like elder care responsibilities.
263
00:21:05,080 --> 00:21:11,480
So it makes it challenging to be in office and present as much as maybe you would like
264
00:21:11,480 --> 00:21:12,480
to be.
265
00:21:12,480 --> 00:21:15,760
There's the there's the old adage, right?
266
00:21:15,760 --> 00:21:17,440
Culture eats strategy for breakfast.
267
00:21:17,440 --> 00:21:22,880
So are we really fixated on our corporate strategy?
268
00:21:22,880 --> 00:21:26,120
Or are we really fixated on our corporate culture?
269
00:21:26,120 --> 00:21:29,080
And I think you can build this remotely.
270
00:21:29,080 --> 00:21:31,640
It is hard.
271
00:21:31,640 --> 00:21:34,000
It might actually be more expensive than what people think.
272
00:21:34,000 --> 00:21:39,280
Like I hope that founders are not choosing hybrid or remote because it's the cheap thing
273
00:21:39,280 --> 00:21:45,840
to do, but that it actually does resonate with the team or individuals or what it is
274
00:21:45,840 --> 00:21:51,840
that that you need to build or why you're building.
275
00:21:51,840 --> 00:21:58,560
You get a lot more accomplished in person, so I do see sort of, you know, community building
276
00:21:58,560 --> 00:22:05,000
this kind of like this culture building happens just a little bit more seamlessly and in person
277
00:22:05,000 --> 00:22:10,080
than it does virtually, but it doesn't mean that it's impossible.
278
00:22:10,080 --> 00:22:16,400
On my experience and kind of our team right now is you have to put effort on both sides.
279
00:22:16,400 --> 00:22:23,440
So we are hybrid, but we have a lot of people that are committed to like, when are we all
280
00:22:23,440 --> 00:22:28,240
going skating in a few weeks together as a team over lunch?
281
00:22:28,240 --> 00:22:34,400
This isn't dictated by the leadership team to say like, you know, 28th of the month is
282
00:22:34,400 --> 00:22:38,880
you know, skating day and we expect you like, no, like it's a real group of friends and
283
00:22:38,880 --> 00:22:43,040
community within the team saying there's literally a skating rink across the street.
284
00:22:43,040 --> 00:22:44,720
Who wants to go on this day?
285
00:22:44,720 --> 00:22:45,720
Yes.
286
00:22:45,720 --> 00:22:47,760
Do this as a team activity.
287
00:22:47,760 --> 00:22:55,080
So there's real effort on that side to build relationships, build trust, build friendships,
288
00:22:55,080 --> 00:23:01,720
get to know one another beyond the code or the deliverable or the tactic that you're
289
00:23:01,720 --> 00:23:02,960
working on.
290
00:23:02,960 --> 00:23:06,800
In the same time, you know, we've also grown remotely.
291
00:23:06,800 --> 00:23:13,840
We have people that are not working within 30 minutes or an hour of downtown Toronto.
292
00:23:13,840 --> 00:23:20,120
How do we maintain that same friendship bond culture virtually?
293
00:23:20,120 --> 00:23:23,520
So you know, there's there's stuff that randomly pops up.
294
00:23:23,520 --> 00:23:27,800
You got to be creative with how you're using Slack.
295
00:23:27,800 --> 00:23:35,440
And I'd say the team does a fairly good job of it, but you need to put effort in into
296
00:23:35,440 --> 00:23:37,360
both sides for sure.
297
00:23:37,360 --> 00:23:41,720
It's something I will actually just share about our team.
298
00:23:41,720 --> 00:23:46,800
We I don't actually know where the name came from, but for whatever reason, on a monthly
299
00:23:46,800 --> 00:23:53,480
basis, they created just a really quick internal application called Timbits.
300
00:23:53,480 --> 00:23:57,280
I don't want any kind of issues with the Tim Horns Corporation.
301
00:23:57,280 --> 00:23:58,280
It's spelled differently.
302
00:23:58,280 --> 00:24:00,480
It just sounds the same.
303
00:24:00,480 --> 00:24:05,880
But on a monthly basis, we just get an email of like, Hey, Mike, you're going to have coffee,
304
00:24:05,880 --> 00:24:11,440
your virtual coffee with Rashad, and you're not allowed to talk about work.
305
00:24:11,440 --> 00:24:14,160
There's some suggestions of things that you could talk about.
306
00:24:14,160 --> 00:24:17,880
And it randomly kind of assigns people on every month.
307
00:24:17,880 --> 00:24:20,360
So this isn't leadership being dictated.
308
00:24:20,360 --> 00:24:23,120
This isn't, you know, the standing meeting.
309
00:24:23,120 --> 00:24:29,400
This isn't, but it does fork force us to break the mold to say, you know what, I haven't
310
00:24:29,400 --> 00:24:33,640
talked to this ETL developer in six months.
311
00:24:33,640 --> 00:24:35,280
We never cross paths.
312
00:24:35,280 --> 00:24:39,640
They're not on, you know, my immediate team.
313
00:24:39,640 --> 00:24:40,640
This is great.
314
00:24:40,640 --> 00:24:41,640
What are you doing as a person?
315
00:24:41,640 --> 00:24:44,920
Did you enjoy the holidays or, you know, how are things going?
316
00:24:44,920 --> 00:24:51,880
So we're using tech and investing sort of both sides of like in person and virtual to
317
00:24:51,880 --> 00:24:55,160
really maintain our team culture.
318
00:24:55,160 --> 00:24:59,600
If you could wave a magic wand and change one thing about Canadian healthcare, what
319
00:24:59,600 --> 00:25:04,760
would you change?
320
00:25:04,760 --> 00:25:13,000
Working in healthcare at times, it feels like, it feels sometimes a little bit like everyone's
321
00:25:13,000 --> 00:25:14,000
at odds.
322
00:25:14,000 --> 00:25:20,280
So, you know, we read a lot of the media and it feels like it's sort of accusatory of
323
00:25:20,280 --> 00:25:25,000
physicians or providers are so slow or something like wait times are egregious because they're
324
00:25:25,000 --> 00:25:26,000
so slow.
325
00:25:26,000 --> 00:25:34,040
If I could wave a magic wand, the one thing I would change is I hope that we come to understanding
326
00:25:34,040 --> 00:25:41,840
that providers and patients actually want the same thing.
327
00:25:41,840 --> 00:25:46,880
We want to take care of our patients and as a patient, you know, I want to get the best
328
00:25:46,880 --> 00:25:50,720
care possible.
329
00:25:50,720 --> 00:25:55,400
What probably your average person doesn't know is all the constraints that are in the
330
00:25:55,400 --> 00:26:02,800
way to deliver, you know, max amount of coverage or the type of quality of care that all of
331
00:26:02,800 --> 00:26:04,280
us want.
332
00:26:04,280 --> 00:26:11,080
And a lot of this is regulatory, a lot of this is data access.
333
00:26:11,080 --> 00:26:13,560
Some of it is human capital.
334
00:26:13,560 --> 00:26:17,880
So I'd hope that I can't change all those things with this wand wand, but the one thing
335
00:26:17,880 --> 00:26:24,040
I would change is just that current, what I feel mindset of that we're at odds like
336
00:26:24,040 --> 00:26:28,760
providers and patients are sort of antagonistic with one another.
337
00:26:28,760 --> 00:26:32,840
I don't feel like it's real and I'd love to wave a wand and just remove that.
338
00:26:32,840 --> 00:26:37,560
Or we all are like, oh yeah, no, they actually want to take care of me and they chose this
339
00:26:37,560 --> 00:26:43,000
profession or you know, you know what, it's fair that this person's really frustrated.
340
00:26:43,000 --> 00:26:47,720
They couldn't get into my family practice for three weeks or four weeks because I'm so
341
00:26:47,720 --> 00:26:52,600
overbooked and now they have pneumonia or they just sat in my emergency department for
342
00:26:52,600 --> 00:26:55,760
12 or 16 hours before they saw me.
343
00:26:55,760 --> 00:26:57,920
And of course, they're really, really frustrated.
344
00:26:57,920 --> 00:27:04,640
So I'd love to sort of remove the animosity so that we could really build back a system
345
00:27:04,640 --> 00:27:07,720
that is reflective of who we are.
346
00:27:07,720 --> 00:27:15,800
If Sequoia or A16Z or General Catalyst sent you an email tomorrow said, Michael, here's
347
00:27:15,800 --> 00:27:17,480
$50 million.
348
00:27:17,480 --> 00:27:19,360
We want you to build something in healthcare.
349
00:27:19,360 --> 00:27:20,360
What would you build?
350
00:27:20,360 --> 00:27:21,360
All right, great.
351
00:27:21,360 --> 00:27:26,640
Well, 50 is not enough money if you're A16Z listening to this.
352
00:27:26,640 --> 00:27:29,720
$500 million, whatever the number is.
353
00:27:29,720 --> 00:27:35,040
So I think, you know, we are starting to see this.
354
00:27:35,040 --> 00:27:39,880
We see some of the potential and the type of quality of data that the Nordic countries
355
00:27:39,880 --> 00:27:41,440
have captured.
356
00:27:41,440 --> 00:27:49,840
And where we're really excited at Unity Health and St. Mike's is yes, AI is everywhere and
357
00:27:49,840 --> 00:27:51,240
this is another AI talk.
358
00:27:51,240 --> 00:27:52,240
I get it.
359
00:27:52,240 --> 00:27:53,920
We're done with that.
360
00:27:53,920 --> 00:27:55,520
We are just at the beginning.
361
00:27:55,520 --> 00:27:59,960
If we're just seeing the tip of the iceberg, we don't see what's below the water.
362
00:27:59,960 --> 00:28:10,160
And the thing that I would build is a real robust multimodal AI company.
363
00:28:10,160 --> 00:28:15,240
Our leader, my boss, Dr. Muhammad Nani, has been talking about this for years.
364
00:28:15,240 --> 00:28:22,120
But imagine a system that can not only read, which is what we have now, right?
365
00:28:22,120 --> 00:28:28,760
A lot of the systems are based on reading text data out of your electronic medical record.
366
00:28:28,760 --> 00:28:36,160
Imagine a future AI that can not only read, but can see because we've built in video feeds.
367
00:28:36,160 --> 00:28:40,200
Imagine a future AI system that can read, see, and hear because we're using auditory
368
00:28:40,200 --> 00:28:46,240
inputs from our emergency department or our surgical bays.
369
00:28:46,240 --> 00:28:51,080
Imagine that it's a system that can feel because it's getting sensor feedback or instrument
370
00:28:51,080 --> 00:28:55,720
feedback from any of the diagnostic machines we're using or any of the surgical devices
371
00:28:55,720 --> 00:28:58,240
that we're using.
372
00:28:58,240 --> 00:29:04,840
That's the type of AI that really feels science fiction and far-fetched, but where I'm getting
373
00:29:04,840 --> 00:29:12,760
excited because we are seeing things like Alpha Fold and the team getting a Nobel Prize
374
00:29:12,760 --> 00:29:15,200
just a few weeks ago.
375
00:29:15,200 --> 00:29:22,680
And that is really, yes, a big, massive data set and a huge feat, but we've really not
376
00:29:22,680 --> 00:29:27,040
tapped into all the data possible.
377
00:29:27,040 --> 00:29:34,800
And if we can get a team like the next DeepMind access to that type of information, and I
378
00:29:34,800 --> 00:29:39,520
think you would be aligned with this, imagine that we can start helping people before they
379
00:29:39,520 --> 00:29:46,800
arrive in clinic that you knew, based on your aura ring and your feedback, I got to actually
380
00:29:46,800 --> 00:29:52,760
get on medication or I need to go to St. Mike's because this is actually pretty severe.
381
00:29:52,760 --> 00:29:57,520
And so the thing I would build is something around multimodal AI.
382
00:29:57,520 --> 00:30:03,080
And I think it's bigger than one medical discipline, one clinical area.
383
00:30:03,080 --> 00:30:07,720
I think it's widespread because medicine is interdisciplinary.
384
00:30:07,720 --> 00:30:14,760
It's going to take in data and data types and feeds from across the care spectrum.
385
00:30:14,760 --> 00:30:21,240
You know, lately we've been hearing that AI, in some circles that AI is getting dumber,
386
00:30:21,240 --> 00:30:26,760
and which is attributed to the phenomenon called drift from what I understand AI just
387
00:30:26,760 --> 00:30:28,840
goes wonky.
388
00:30:28,840 --> 00:30:30,280
How do you control for that?
389
00:30:30,280 --> 00:30:32,160
How do you fix that?
390
00:30:32,160 --> 00:30:44,720
Yeah, I think we are seeing that drift is not a guarantee from at least my reading and
391
00:30:44,720 --> 00:30:45,720
understanding.
392
00:30:45,720 --> 00:30:47,520
So it may or may not occur.
393
00:30:47,520 --> 00:30:52,200
And that's sort of the interesting phenomena to me of like, why is this or when is this
394
00:30:52,200 --> 00:30:55,880
not happening?
395
00:30:55,880 --> 00:31:03,120
Some of our models which are not LMS are just the R machine learning and some rules based
396
00:31:03,120 --> 00:31:05,000
systems, they're not getting dumber.
397
00:31:05,000 --> 00:31:07,440
The accuracy is staying the same.
398
00:31:07,440 --> 00:31:14,600
But the part that I think maybe we need to keep in mind is our current context does change.
399
00:31:14,600 --> 00:31:16,240
So I'll give you an example.
400
00:31:16,240 --> 00:31:21,040
We've got a tool that predicts volumes into our emergency department at St. Mike's and
401
00:31:21,040 --> 00:31:22,040
St. Joe's.
402
00:31:22,040 --> 00:31:28,640
It's 48 to 72 hours in advance and it's very, very accurate, over 90% accurate, I think
403
00:31:28,640 --> 00:31:30,760
over 95% accurate.
404
00:31:30,760 --> 00:31:37,840
The part that we've done that's different than some other areas is we're able to predict
405
00:31:37,840 --> 00:31:41,280
severity of cases and some key areas for us.
406
00:31:41,280 --> 00:31:49,400
So we'll know, okay, you know, on Monday morning, we are going to have 30 cases, 28 of them are
407
00:31:49,400 --> 00:31:53,760
going to be pretty easy, two of them are going to be difficult and five of them are going
408
00:31:53,760 --> 00:31:55,840
to be mental health cases.
409
00:31:55,840 --> 00:32:00,880
What is this tool trained on or where is it getting its data from just in general?
410
00:32:00,880 --> 00:32:06,200
Yeah, so this is, it's a time series model.
411
00:32:06,200 --> 00:32:11,680
And so we're doing inputs of sort of our historical data at our emergency departments of the types
412
00:32:11,680 --> 00:32:12,680
of visits.
413
00:32:12,680 --> 00:32:17,400
And there's a lot of literature, I think going back to the 70s of that, emergency department
414
00:32:17,400 --> 00:32:21,720
visits are very predictable and there's seasonality associated with it.
415
00:32:21,720 --> 00:32:27,480
As Canadians, we should understand that there's cold and flu season for, for a reason.
416
00:32:27,480 --> 00:32:30,520
For our hospital too, there's also trauma care season.
417
00:32:30,520 --> 00:32:36,620
So please wear your helmets when you're riding your bike or going skating.
418
00:32:36,620 --> 00:32:40,640
So we see these natural spikes and occurrences and they do repeat themselves on an annual
419
00:32:40,640 --> 00:32:41,900
basis.
420
00:32:41,900 --> 00:32:51,880
The other key input for us too, actually, is we input weather and we input cultural events
421
00:32:51,880 --> 00:32:53,520
in the city.
422
00:32:53,520 --> 00:32:58,200
So really, really high likelihood, but we know, you know, if there's a hockey game and
423
00:32:58,200 --> 00:33:03,320
freezing rain and it's a Saturday night during cold and flu season, it's probably going to
424
00:33:03,320 --> 00:33:05,480
be a bad night in our emergency department.
425
00:33:05,480 --> 00:33:11,080
And when you say you input cultural events, like what is actually happening?
426
00:33:11,080 --> 00:33:17,160
Yeah, so we have basically like a listening type bot that there's a few different sites
427
00:33:17,160 --> 00:33:19,640
that's looking for posted dates.
428
00:33:19,640 --> 00:33:25,680
So hockey games is, it's one of the big ones basketball games.
429
00:33:25,680 --> 00:33:30,640
We you know, if there's parades and stuff like that, there's just more slips, trips,
430
00:33:30,640 --> 00:33:32,000
bumps and falls.
431
00:33:32,000 --> 00:33:34,120
And so therefore, where do they arrive?
432
00:33:34,120 --> 00:33:36,480
They arrive at downtown hospitals.
433
00:33:36,480 --> 00:33:42,280
And it's able to kind of say a hockey game will have one extra patient, whereas a Taylor
434
00:33:42,280 --> 00:33:45,480
Swift concert might have 100 extra patients or.
435
00:33:45,480 --> 00:33:52,360
Yeah, I don't know the volumes and how that, how those two sort of things equate.
436
00:33:52,360 --> 00:33:54,720
I will tell you the thing that we missed.
437
00:33:54,720 --> 00:34:00,160
I don't know if the listening tool didn't have this, but the thing where we had many
438
00:34:00,160 --> 00:34:05,360
more people in the emergency department than what we were expecting was when our world
439
00:34:05,360 --> 00:34:10,160
championship Toronto Raptors had their parade.
440
00:34:10,160 --> 00:34:16,520
And we, you know, I don't think the city was ready for that celebration.
441
00:34:16,520 --> 00:34:20,720
But the emergency department, it was like we weren't ready, but we saw many more people
442
00:34:20,720 --> 00:34:24,080
than what we were anticipating that day.
443
00:34:24,080 --> 00:34:26,680
So that's one of the, one of the misses.
444
00:34:26,680 --> 00:34:33,240
But the, yeah, I don't know that we take into account the volume of the event, but just
445
00:34:33,240 --> 00:34:38,680
certain events like, you know, marathons and stuff like that, the people do get hurt and
446
00:34:38,680 --> 00:34:42,480
they do end up in our emergency department.
447
00:34:42,480 --> 00:34:48,360
Do you think we should mandate explainability for AI as some of our medical societies have
448
00:34:48,360 --> 00:34:54,160
come out with and said, you know, if you're using an AI, you need to understand at some
449
00:34:54,160 --> 00:35:00,440
level how it works, but AI, you know, is a black box for the most part.
450
00:35:00,440 --> 00:35:01,440
Yeah.
451
00:35:01,440 --> 00:35:07,920
So I think explainability has caught on somewhat unfortunately.
452
00:35:07,920 --> 00:35:11,360
And so there's a few things to be the backup and say about this.
453
00:35:11,360 --> 00:35:15,160
So no, I don't think that we should mandate explainability and feel like also that we've
454
00:35:15,160 --> 00:35:16,680
solved the problem.
455
00:35:16,680 --> 00:35:23,720
I feel like explainability is caught on as a surrogate for trust and safety.
456
00:35:23,720 --> 00:35:29,080
So if we can understand something is kind of the argument, oh, then therefore it is trustworthy
457
00:35:29,080 --> 00:35:31,000
or then therefore it is safe.
458
00:35:31,000 --> 00:35:32,000
And that's actually not true.
459
00:35:32,000 --> 00:35:36,920
You can understand something and it could still not be trustworthy and it could still
460
00:35:36,920 --> 00:35:38,440
not be safe.
461
00:35:38,440 --> 00:35:43,360
The other example, you know, I've heard many people sort of say, you know, there's actually
462
00:35:43,360 --> 00:35:53,120
a lot of really common medications that all of us use regularly that we don't understand
463
00:35:53,120 --> 00:35:55,480
that we can't explain.
464
00:35:55,480 --> 00:35:59,360
You can ask your average physician or nurse, what is the mechanism of action for a CETA
465
00:35:59,360 --> 00:36:02,680
medifin and they're probably going to get it wrong.
466
00:36:02,680 --> 00:36:08,760
But yet it's, you know, one of the most widely used pharmaceuticals in the world.
467
00:36:08,760 --> 00:36:13,280
We all drive cars or get on trains and buses, but we don't necessarily understand, you know,
468
00:36:13,280 --> 00:36:17,880
how does a V8 or V12 engine works anymore?
469
00:36:17,880 --> 00:36:21,240
What is the difference between combustion or electric engines?
470
00:36:21,240 --> 00:36:26,520
So I don't think that explainability takes us to the place that we want to go.
471
00:36:26,520 --> 00:36:30,320
I think it just sort of caught on of like, oh, you know, if I understand more, I'm going
472
00:36:30,320 --> 00:36:32,880
to feel safer.
473
00:36:32,880 --> 00:36:37,760
And the reality with AI is you might not understand something, but still it might be really, really
474
00:36:37,760 --> 00:36:39,280
unsafe.
475
00:36:39,280 --> 00:36:47,600
So the part where I'm interested is what are the monitors and evaluations that are in place.
476
00:36:47,600 --> 00:36:50,320
And I don't know how we necessarily regulate that.
477
00:36:50,320 --> 00:36:52,640
That's not my space to dictate.
478
00:36:52,640 --> 00:36:59,800
But what I would want to see is for critical systems that have an impact in hurting someone,
479
00:36:59,800 --> 00:37:05,560
that they're monitored, and that there's some type of regular evaluation in place.
480
00:37:05,560 --> 00:37:11,520
That's what would make me feel safer and that we're in a better spot.
481
00:37:11,520 --> 00:37:19,120
Throughout history, we have been the smartest being on earth as far as we know.
482
00:37:19,120 --> 00:37:24,560
And now we're creating something which is more creative and smarter than us.
483
00:37:24,560 --> 00:37:26,880
And I think this scares us.
484
00:37:26,880 --> 00:37:34,040
And I think this is where this rise of explainability comes from is, okay, if I can explain it,
485
00:37:34,040 --> 00:37:36,600
I'm smarter than it.
486
00:37:36,600 --> 00:37:43,720
And I think that that's what's driving these rules and regulations.
487
00:37:43,720 --> 00:37:50,160
You sit at an intersection of industry and healthcare.
488
00:37:50,160 --> 00:37:59,800
What advice would you give to hospitals who are looking to be more innovative and incorporate
489
00:37:59,800 --> 00:38:03,840
more AI and startups into their ecosystem?
490
00:38:03,840 --> 00:38:10,160
And what are some things they do wrong?
491
00:38:10,160 --> 00:38:18,480
So I think it really comes down to identifying the problem and picking the right problem.
492
00:38:18,480 --> 00:38:24,360
And there's many different ways that you can go about that.
493
00:38:24,360 --> 00:38:28,560
What I see, whether you're building or you're buying something, like you're going to partner
494
00:38:28,560 --> 00:38:34,560
with a startup or what have you, innovation is this big, sexy thing.
495
00:38:34,560 --> 00:38:39,600
And a lot of people on the leadership team get involved in these meetings and issue the
496
00:38:39,600 --> 00:38:41,800
RFP, what have you.
497
00:38:41,800 --> 00:38:47,680
And I'll just be honest, I don't know many nurses directly.
498
00:38:47,680 --> 00:38:52,160
I'm not funded or supported by any sort of nursing union or federation.
499
00:38:52,160 --> 00:38:58,360
But you go through these processes and typically nurses are not included whatsoever.
500
00:38:58,360 --> 00:39:01,360
And it's such a huge miss for acute care.
501
00:39:01,360 --> 00:39:07,040
I can't speak to clinical practice specifically.
502
00:39:07,040 --> 00:39:12,800
But for acute care, where it's like something 80% or like 70% of bedside care is delivered
503
00:39:12,800 --> 00:39:16,520
by nurses, but they're not part of the innovation RFP process.
504
00:39:16,520 --> 00:39:22,760
Or they don't have a mechanism to share their challenges or pain points.
505
00:39:22,760 --> 00:39:30,600
So that for me is like a big miss of not including everyone within that practice as part of the
506
00:39:30,600 --> 00:39:36,880
journey, identifying the right end users as well.
507
00:39:36,880 --> 00:39:43,400
So we're still seeing very much a top down model.
508
00:39:43,400 --> 00:39:50,160
And I think we need to be honest, our innovation record in healthcare, but across industries
509
00:39:50,160 --> 00:39:54,220
in Canada is one of the worst in the world.
510
00:39:54,220 --> 00:40:02,320
So we have the lowest corporate R&D expenditure on technology in the G8.
511
00:40:02,320 --> 00:40:06,280
So we find ourselves below Italy.
512
00:40:06,280 --> 00:40:10,160
So that's not something that we might kind of like really conceptualize and think about.
513
00:40:10,160 --> 00:40:17,200
This isn't healthcare being less than average at its ability to innovate.
514
00:40:17,200 --> 00:40:23,160
We're seeing this in financial services, in manufacturing, in forestry, in education,
515
00:40:23,160 --> 00:40:28,520
in any industry in Canada, we find ourselves in the last place of the G8.
516
00:40:28,520 --> 00:40:33,920
So there's kind of this thing, like if you do something wrong once, do you repeat that
517
00:40:33,920 --> 00:40:35,960
pattern or do you actually change it?
518
00:40:35,960 --> 00:40:41,640
We find ourselves decades into this journey of we keep repeating the same pattern.
519
00:40:41,640 --> 00:40:44,880
And what I would say is we need to break the model.
520
00:40:44,880 --> 00:40:46,280
We need to try new things.
521
00:40:46,280 --> 00:40:53,600
Probably the same model doesn't work for each organization or each individual, but not including
522
00:40:53,600 --> 00:40:59,800
your people in that process, in that journey is a huge, huge miss.
523
00:40:59,800 --> 00:41:07,520
My second thing that probably we'll get some hate mail for this one, but I've said it before,
524
00:41:07,520 --> 00:41:09,120
so it's okay.
525
00:41:09,120 --> 00:41:19,280
The second fallacy that I think organizations fall into is a very long, protracted, expensive
526
00:41:19,280 --> 00:41:21,280
governance process.
527
00:41:21,280 --> 00:41:27,080
Okay, so we are going to innovate XYZ.
528
00:41:27,080 --> 00:41:33,320
So we need to talk to privacy, we need to talk to legal, and we need to talk to IT security,
529
00:41:33,320 --> 00:41:39,880
we need to talk to ethics, and we probably need to hire a consultant and, you know, met
530
00:41:39,880 --> 00:41:46,600
with one of my colleagues and they're frustrated, but they're into two plus years of working
531
00:41:46,600 --> 00:41:50,800
on an AI governance framework for their organization.
532
00:41:50,800 --> 00:41:53,680
It's the most beautiful framework you will ever see.
533
00:41:53,680 --> 00:41:59,480
I hope they get it framed, but where we stand today after the tens of thousands of hours
534
00:41:59,480 --> 00:42:06,680
that have been invested into that thing is the organization has not deployed any AI solutions.
535
00:42:06,680 --> 00:42:12,600
So we're now in 2025 and they're not using AI after all that money.
536
00:42:12,600 --> 00:42:14,960
So I wouldn't get hung up.
537
00:42:14,960 --> 00:42:19,560
Yes, you need governance, but you need to be lean, smart, and action oriented of how you're
538
00:42:19,560 --> 00:42:21,640
going through that process.
539
00:42:21,640 --> 00:42:23,920
And yes, leadership needs to be engaged.
540
00:42:23,920 --> 00:42:28,920
We would have that as part of like our strategy framework or our diamondy.
541
00:42:28,920 --> 00:42:34,120
Yes, leadership needs to be aligned, but leadership doesn't need to lead and be the only decision
542
00:42:34,120 --> 00:42:36,680
maker in an innovation process.
543
00:42:36,680 --> 00:42:42,480
We need to keep our people and our end users engaged and part of that process as well.
544
00:42:42,480 --> 00:42:50,840
Yeah, I think hospitals need to have an innovation department which has control over a few beds
545
00:42:50,840 --> 00:42:59,280
in a marriage or in a floor and has full and has a budget and has the ability to deploy
546
00:42:59,280 --> 00:43:03,760
that budget without any oversight from leadership.
547
00:43:03,760 --> 00:43:11,600
I think there needs to be a distinct department which is separate from hospital governance,
548
00:43:11,600 --> 00:43:15,960
legal leadership that can, you know, incorporate these projects.
549
00:43:15,960 --> 00:43:16,960
Yeah, I completely agree.
550
00:43:16,960 --> 00:43:24,760
And I mean, like your audience is mostly in healthcare, so they will get this, but we
551
00:43:24,760 --> 00:43:27,040
see this in other areas as well.
552
00:43:27,040 --> 00:43:31,520
So I spent some time in academia as I mentioned.
553
00:43:31,520 --> 00:43:37,320
If we look at our org chart, how come we don't have a chief innovation officer or a chief
554
00:43:37,320 --> 00:43:38,480
technology officer?
555
00:43:38,480 --> 00:43:44,520
Like, isn't that strange that like some of these networks and systems are multi-million,
556
00:43:44,520 --> 00:43:49,320
sometimes billion dollar organizations and we don't have a chief innovation or chief
557
00:43:49,320 --> 00:43:51,000
technology officer?
558
00:43:51,000 --> 00:43:55,240
We also sometimes don't have a chief marketing officer or chief revenue officer either.
559
00:43:55,240 --> 00:43:59,040
And this goes for major academic health science networks.
560
00:43:59,040 --> 00:44:03,120
This goes for many of the universities and colleges in the country.
561
00:44:03,120 --> 00:44:08,920
So, you know, wearing my like business school hat of like, why do we find ourselves in the
562
00:44:08,920 --> 00:44:10,560
position that we are?
563
00:44:10,560 --> 00:44:15,720
Yeah, you know, we can tweak some things and we can blame economic policy or hate on whatever
564
00:44:15,720 --> 00:44:22,320
political party we want or individual, but structurally, we've made some really weird
565
00:44:22,320 --> 00:44:23,320
decisions, right?
566
00:44:23,320 --> 00:44:33,200
Like, if we want to see more innovation commercialized out of our universities and academic hospitals,
567
00:44:33,200 --> 00:44:38,360
how come there's not a chief commercialization officer at every university and hospital?
568
00:44:38,360 --> 00:44:44,600
If we want our hospitals to, you know, be global brands like the MIT, the Harford, the
569
00:44:44,600 --> 00:44:49,000
Stanford, the Oxford, it came from U of T and Ivy that have great brands.
570
00:44:49,000 --> 00:44:53,800
But how come we don't have a chief marketing and revenue officer at every university and
571
00:44:53,800 --> 00:44:54,800
hospital?
572
00:44:54,800 --> 00:45:01,240
So, there's things where structurally, we find ourselves where we are because we've made
573
00:45:01,240 --> 00:45:04,000
the decision, we've made the bed that we're lying in.
574
00:45:04,000 --> 00:45:07,800
And that's things that could change, you know, to your point, can you appoint someone?
575
00:45:07,800 --> 00:45:14,040
Can you give them a mandate money and people to go and change the way that care is delivered?
576
00:45:14,040 --> 00:45:15,040
Absolutely.
577
00:45:15,040 --> 00:45:19,480
There's nothing that's holding us back except our own ambition.
578
00:45:19,480 --> 00:45:21,080
Last question, Michael.
579
00:45:21,080 --> 00:45:26,760
If you could go back and talk to yourself 20 years ago, what advice would you give him?
580
00:45:26,760 --> 00:45:31,040
You have no idea what's coming, but take a breath, it's all going to be okay.
581
00:45:31,040 --> 00:45:32,040
Awesome.
582
00:45:32,040 --> 00:45:33,040
Thank you, Michael.
583
00:45:33,040 --> 00:45:34,560
Yeah, great to speak with you.
584
00:45:34,560 --> 00:45:41,560
Thank you.
