1
00:00:00,500 –> 00:00:02,400
MOHAN: Welcome, everybody, to–

2
00:00:02,400 –> 00:00:06,360
this is the third webinar
in the four part series.

3
00:00:06,360 –> 00:00:08,550
The way we’ve laid
out these webinars,

4
00:00:08,550 –> 00:00:11,520
each one progressively
focuses on a different topic.

5
00:00:11,520 –> 00:00:15,990
So webinar one focused
on why the EHR is not

6
00:00:15,990 –> 00:00:19,140
built to optimize scheduling,
and why the mathematics of it

7
00:00:19,140 –> 00:00:21,270
are harder than one might think.

8
00:00:21,270 –> 00:00:23,970
Webinar two focused
on, how do you actually

9
00:00:23,970 –> 00:00:25,620
optimize a schedule?

10
00:00:25,620 –> 00:00:29,790
Today’s webinar
will focus on how

11
00:00:29,790 –> 00:00:32,729
you shape the future performance
of your infusion center.

12
00:00:32,729 –> 00:00:35,790
And the next one will
focus on the diagnostics.

13
00:00:35,790 –> 00:00:38,040
Just in the interest
of the folks

14
00:00:38,040 –> 00:00:40,010
who did not attend
the first two,

15
00:00:40,010 –> 00:00:41,610
I’ll quickly do a refresher.

16
00:00:41,610 –> 00:00:43,800
So within the first
10 or 15 minutes

17
00:00:43,800 –> 00:00:46,930
we’ll cover all of
the prior topics.

18
00:00:46,930 –> 00:00:49,260
Those topics are available
in greater detail

19
00:00:49,260 –> 00:00:51,420
in the recorded webinar,
so feel free to go

20
00:00:51,420 –> 00:00:54,510
download those and watch
them in greater detail.

21
00:00:54,510 –> 00:00:55,410
OK.

22
00:00:55,410 –> 00:00:58,020
So today we will start with
a quick introduction of who

23
00:00:58,020 –> 00:01:02,790
we are, then spend five minutes
in total talking about why EHRs

24
00:01:02,790 –> 00:01:04,800
are not built to
do this for you,

25
00:01:04,800 –> 00:01:08,070
talk about infusion
optimization in five minutes,

26
00:01:08,070 –> 00:01:11,420
and then spend the bulk of our
time talking about planning.

27
00:01:11,420 –> 00:01:12,210
OK.

28
00:01:12,210 –> 00:01:14,560
So who is LeanTaaS
and what do we do?

29
00:01:14,560 –> 00:01:16,320
We are a Silicon Valley
software company.

30
00:01:16,320 –> 00:01:20,520
And our focus is on using
mathematics and software

31
00:01:20,520 –> 00:01:24,150
to unlock capacity in hospitals.

32
00:01:24,150 –> 00:01:27,540
And the magic of
unlocking the capacity

33
00:01:27,540 –> 00:01:29,700
is that many good things follow.

34
00:01:29,700 –> 00:01:32,370
Patient access improves,
meaning it’s a shorter lead

35
00:01:32,370 –> 00:01:34,400
time to future appointments.

36
00:01:34,400 –> 00:01:36,870
Wait time for patients
go down, which

37
00:01:36,870 –> 00:01:39,630
is the biggest source of
patient dissatisfaction,

38
00:01:39,630 –> 00:01:44,070
and so having the capacity to
require patients to wait less

39
00:01:44,070 –> 00:01:45,960
is a very good thing.

40
00:01:45,960 –> 00:01:47,980
Both operating costs
and capital costs

41
00:01:47,980 –> 00:01:51,750
improve, because the ability
to use the staff and the assets

42
00:01:51,750 –> 00:01:54,930
more productively just
results in an overall lower

43
00:01:54,930 –> 00:01:56,190
cost envelope.

44
00:01:56,190 –> 00:02:00,720
And finally, to the extent that
the capacity unlocking allows

45
00:02:00,720 –> 00:02:02,120
you to see more patients.

46
00:02:02,120 –> 00:02:05,190
There is a revenue
uplift from that as well.

47
00:02:05,190 –> 00:02:08,789
The two products that we
have commercially out there

48
00:02:08,789 –> 00:02:11,039
are Infusion and
Operating Rooms.

49
00:02:11,039 –> 00:02:13,200
And we are in the
middle of working

50
00:02:13,200 –> 00:02:16,940
with partners on building
the next set of products.

51
00:02:16,940 –> 00:02:20,310
So for example, we are working
with 20 oncologists each

52
00:02:20,310 –> 00:02:22,860
at Memorial Sloan Kettering
in New York and M.D.

53
00:02:22,860 –> 00:02:26,130
Anderson in Houston, to
build our oncology template

54
00:02:26,130 –> 00:02:29,370
optimization for the providers,
so that the wait times

55
00:02:29,370 –> 00:02:31,230
and clinics can get reduced.

56
00:02:31,230 –> 00:02:33,600
Similarly, we’ve got
parallel initiatives

57
00:02:33,600 –> 00:02:37,120
on imaging inpatient
beds and labs.

58
00:02:37,120 –> 00:02:40,540
In terms of who we work with,
these are the leading health

59
00:02:40,540 –> 00:02:43,360
systems across the
country, and they

60
00:02:43,360 –> 00:02:47,080
span the gamut from academic
medical centers like Stanford

61
00:02:47,080 –> 00:02:49,960
and UCSF, and Duke
and Emory, and UPenn,

62
00:02:49,960 –> 00:02:52,930
to iconic institutions
like M.D. Anderson,

63
00:02:52,930 –> 00:02:57,640
Sloan Kettering, and Johns
Hopkins, to regional hospitals

64
00:02:57,640 –> 00:02:58,870
and cancer centers as well.

65
00:03:02,660 –> 00:03:04,580
The fact that
infusion scheduling

66
00:03:04,580 –> 00:03:08,360
is a pinpoint felt by
many, many cancer centers

67
00:03:08,360 –> 00:03:12,620
is evident in the
fact that from 0 or 1

68
00:03:12,620 –> 00:03:16,880
infusion center in early 2015,
we’re now running 113 infusion

69
00:03:16,880 –> 00:03:19,670
centers with nearly
3,000 shares, based

70
00:03:19,670 –> 00:03:21,630
on the optimization algorithms.

71
00:03:21,630 –> 00:03:25,450
So that’s us in a nutshell.

72
00:03:25,450 –> 00:03:27,890
To give you a quick
refresher on why

73
00:03:27,890 –> 00:03:31,910
is it that the EHR is
not built to optimize,

74
00:03:31,910 –> 00:03:34,050
there are three big
reasons for this.

75
00:03:34,050 –> 00:03:38,690
The first is, as you look
around any health system,

76
00:03:38,690 –> 00:03:42,740
there is a tendency to rely on
a grid-based schedule, which

77
00:03:42,740 –> 00:03:46,370
is in point one, where the
assets are laid across the top

78
00:03:46,370 –> 00:03:48,860
and times of day down the left.

79
00:03:48,860 –> 00:03:51,950
The assets could be chairs,
rooms, providers, imaging

80
00:03:51,950 –> 00:03:54,077
machines, whatever the asset is.

81
00:03:54,077 –> 00:03:55,910
And so when John Smith
needs an 8:00 to 9:00

82
00:03:55,910 –> 00:03:59,300
appointment, someone somewhere
colors in 8:00 to 9:00,

83
00:03:59,300 –> 00:04:02,390
and puts down John Smith’s
name or MRN number,

84
00:04:02,390 –> 00:04:05,090
this is on spreadsheets,
on whiteboards,

85
00:04:05,090 –> 00:04:08,180
on snap boards in
the EHR, et cetera.

86
00:04:08,180 –> 00:04:10,470
This just does not work.

87
00:04:10,470 –> 00:04:13,490
The reason is grid-based
scheduling works

88
00:04:13,490 –> 00:04:15,950
if you’re scheduling something
that is deterministic,

89
00:04:15,950 –> 00:04:18,019
meaning the start
and the end time

90
00:04:18,019 –> 00:04:20,750
is known at the time of
making the appointment,

91
00:04:20,750 –> 00:04:24,110
as is true for tennis
courts and spa treatments.

92
00:04:24,110 –> 00:04:25,970
Medical appointments
and infusion treatments

93
00:04:25,970 –> 00:04:28,710
are stochastic, meaning they
are random and highly variable.

94
00:04:28,710 –> 00:04:30,710
The start time and the
end time doesn’t work out

95
00:04:30,710 –> 00:04:33,820
as you thought,
and therefore using

96
00:04:33,820 –> 00:04:36,770
a deterministic scheduling
framework, like a grid,

97
00:04:36,770 –> 00:04:39,860
to schedule a stochastic thing
like an infusion appointment,

98
00:04:39,860 –> 00:04:42,080
is just flat out
mathematically wrong.

99
00:04:42,080 –> 00:04:44,930
This is why the grid looks
great on the previous night

100
00:04:44,930 –> 00:04:47,420
and the day never
works out as planned.

101
00:04:47,420 –> 00:04:50,070
The second reason,
the mathematics of EHR

102
00:04:50,070 –> 00:04:52,520
is simply not robust enough.

103
00:04:52,520 –> 00:04:55,880
Every EHR, and therefore,
every health system,

104
00:04:55,880 –> 00:04:59,120
follows a first come, first
schedule sort of a discipline

105
00:04:59,120 –> 00:05:00,590
in booking appointments.

106
00:05:00,590 –> 00:05:03,770
Meaning, if you call and ask
for an appointment seven months

107
00:05:03,770 –> 00:05:05,930
into the future,
you typically will

108
00:05:05,930 –> 00:05:08,560
be told the calendar is
open, pick a spot, any spot.

109
00:05:08,560 –> 00:05:10,760
Here are our hours of operation.

110
00:05:10,760 –> 00:05:13,140
That looks nice and
patient centric.

111
00:05:13,140 –> 00:05:16,370
It’s wrong because nothing
mathematically balanced

112
00:05:16,370 –> 00:05:17,420
the load.

113
00:05:17,420 –> 00:05:19,250
If you’re going to
balance the load,

114
00:05:19,250 –> 00:05:21,500
then continuously
throughout the day

115
00:05:21,500 –> 00:05:23,720
you have to balance
the number of people

116
00:05:23,720 –> 00:05:26,990
just getting started, with the
number of people just leaving,

117
00:05:26,990 –> 00:05:28,910
and the number of
people in between.

118
00:05:28,910 –> 00:05:32,120
That’s the way you create a
balance profile, first come,

119
00:05:32,120 –> 00:05:35,610
first schedule just does
not allow that to happen.

120
00:05:35,610 –> 00:05:37,580
And the third thing
that’s mathematically,

121
00:05:37,580 –> 00:05:41,240
completely inadequate
in EHRs, is this notion

122
00:05:41,240 –> 00:05:43,160
of a connecting appointment.

123
00:05:43,160 –> 00:05:45,590
Almost everything
in a health system

124
00:05:45,590 –> 00:05:48,500
is a connected series of
appointments, labs followed

125
00:05:48,500 –> 00:05:49,880
by clinics, followed
by infusion,

126
00:05:49,880 –> 00:05:51,770
followed by radiation oncology.

127
00:05:51,770 –> 00:05:54,470
When connected
schedules are built,

128
00:05:54,470 –> 00:05:59,627
they have to first off ensure
that the earlier segments

129
00:05:59,627 –> 00:06:01,460
operate on time, just
like your first flight

130
00:06:01,460 –> 00:06:02,877
has to be on time
in order for you

131
00:06:02,877 –> 00:06:04,280
to make your connecting flight.

132
00:06:04,280 –> 00:06:06,020
That doesn’t often happen.

133
00:06:06,020 –> 00:06:08,600
And the second thing is,
these connected schedules

134
00:06:08,600 –> 00:06:10,250
are built one person at a time.

135
00:06:10,250 –> 00:06:11,803
So John Smith is
given the 8 o’clock

136
00:06:11,803 –> 00:06:13,970
appointment at the doc and
the 9 o’clock appointment

137
00:06:13,970 –> 00:06:15,890
at the infusion,
and then along comes

138
00:06:15,890 –> 00:06:17,930
Jane Doe and the
same thing happens.

139
00:06:17,930 –> 00:06:19,910
Connecting schedules
need to be built

140
00:06:19,910 –> 00:06:22,340
optimizing the system
as a whole, not one

141
00:06:22,340 –> 00:06:23,400
person at a time.

142
00:06:23,400 –> 00:06:25,317
So when you think about
how connecting flights

143
00:06:25,317 –> 00:06:27,275
are put together,
they don’t sort out

144
00:06:27,275 –> 00:06:29,150
the departure time of
the second flight based

145
00:06:29,150 –> 00:06:30,530
on one passenger at a time.

146
00:06:30,530 –> 00:06:32,280
They optimize the whole system.

147
00:06:32,280 –> 00:06:34,460
So for these reasons,
the mathematics

148
00:06:34,460 –> 00:06:37,920
is just simply not adequate.

149
00:06:37,920 –> 00:06:43,760
Now, infusion at the surface
looks like it should be simple.

150
00:06:43,760 –> 00:06:45,530
It’s a person
sitting in a chair,

151
00:06:45,530 –> 00:06:49,040
getting infused for some
number of hours, from one

152
00:06:49,040 –> 00:06:50,120
to nine hours.

153
00:06:50,120 –> 00:06:54,140
And so on the surface it says,
how hard can it possibly be?

154
00:06:54,140 –> 00:06:56,570
Well, it turns out it’s
actually very, very hard.

155
00:06:56,570 –> 00:07:00,590
Even if you took five types
of appointments, one hour, two

156
00:07:00,590 –> 00:07:04,010
hour, three to five, six
to eight, and nine plus,

157
00:07:04,010 –> 00:07:08,150
assume a modest sized infusion
center, 20, 25 chairs that

158
00:07:08,150 –> 00:07:10,480
sees up to 70 patients a day.

159
00:07:10,480 –> 00:07:15,500
With this duration mix, followed
by how many possible slots can

160
00:07:15,500 –> 00:07:17,270
there be, if you can
offer up appointments

161
00:07:17,270 –> 00:07:20,660
at 10 minute intervals
7:00, 7:10, 7:20, and so on,

162
00:07:20,660 –> 00:07:22,650
you’ll have 64 slots.

163
00:07:22,650 –> 00:07:24,590
And if you could seat
four patients at a time,

164
00:07:24,590 –> 00:07:27,110
that’s 256 possible slots.

165
00:07:27,110 –> 00:07:32,600
Mathematically, that’s a number
with 105 zeros behind it.

166
00:07:32,600 –> 00:07:35,270
That’s the number of
permutations and combinations

167
00:07:35,270 –> 00:07:38,330
for how you could
build that schedule.

168
00:07:38,330 –> 00:07:41,150
A number with 105 zeros
behind it is staggering.

169
00:07:41,150 –> 00:07:44,000
Just to put this in
context, the odds

170
00:07:44,000 –> 00:07:47,240
of winning the Mega Millions
is 1 in 259 million,

171
00:07:47,240 –> 00:07:49,910
the odds of winning
Powerball is 1 in 292,

172
00:07:49,910 –> 00:07:51,650
if you wanted to win
both of those, one

173
00:07:51,650 –> 00:07:55,230
after the other, that’s a
number with 15 zeros behind it.

174
00:07:55,230 –> 00:07:58,220
So you’d have to win both of
them, seven times in a row,

175
00:07:58,220 –> 00:08:00,820
in order to get the infusion
schedule right for just one

176
00:08:00,820 –> 00:08:01,420
day.

177
00:08:01,420 –> 00:08:03,880
That’s how mathematically
overwhelming it is.

178
00:08:03,880 –> 00:08:09,670
And so expecting schedulers and
nurse managers and other clinic

179
00:08:09,670 –> 00:08:13,360
folks to just pick a number,
pick a slot on the calendar,

180
00:08:13,360 –> 00:08:17,470
and get it even remotely right
is just not going to happen.

181
00:08:17,470 –> 00:08:19,870
After looking at all
these permutations, what

182
00:08:19,870 –> 00:08:22,300
makes it harder is
there are real life

183
00:08:22,300 –> 00:08:24,010
operational constraints.

184
00:08:24,010 –> 00:08:28,270
The volume and mix for each
day is unique to a day of week.

185
00:08:28,270 –> 00:08:29,880
The nurse availability
and workload,

186
00:08:29,880 –> 00:08:32,260
which we’ll spend more
time on, is unique.

187
00:08:32,260 –> 00:08:35,530
The chair availability
also depends.

188
00:08:35,530 –> 00:08:38,650
And finally, there’s lots
of expected and unexpected

189
00:08:38,650 –> 00:08:40,277
variability of
clinics running late,

190
00:08:40,277 –> 00:08:42,610
which you could have predicted,
and clinics running late

191
00:08:42,610 –> 00:08:45,068
or patients running late that
you could not have predicted.

192
00:08:45,068 –> 00:08:48,130
And so all of these have
to be taken into account.

193
00:08:48,130 –> 00:08:50,680
Because of the maps
not being sufficient,

194
00:08:50,680 –> 00:08:54,220
in most infusion centers,
the day plays out like this.

195
00:08:54,220 –> 00:08:57,880
In a peaky profile, where
patients arrive roughly

196
00:08:57,880 –> 00:09:00,070
in the order of
their appointment

197
00:09:00,070 –> 00:09:02,440
with varying treatment
lengths needed,

198
00:09:02,440 –> 00:09:04,930
the charge nurse tries to
put them in the right part,

199
00:09:04,930 –> 00:09:06,880
in the right chair,
with the right nurse.

200
00:09:06,880 –> 00:09:11,360
And the game of Tetris as
it unfolds is a losing hand.

201
00:09:11,360 –> 00:09:14,640
The reason this matters is
the duration of the peak

202
00:09:14,640 –> 00:09:15,940
is only three or four hours.

203
00:09:15,940 –> 00:09:18,598
And the duration of a
nursing shift is eight or 10.

204
00:09:18,598 –> 00:09:20,140
Which means, right
off the bat you’re

205
00:09:20,140 –> 00:09:22,420
confronted with a bad choice.

206
00:09:22,420 –> 00:09:24,370
Should you staff for the peak?

207
00:09:24,370 –> 00:09:27,010
In which case you are
overstaffed before and after.

208
00:09:27,010 –> 00:09:28,930
Or should you staff
for the average?

209
00:09:28,930 –> 00:09:31,750
In which case, at the peak,
right when you need it most,

210
00:09:31,750 –> 00:09:33,340
you are understaffed.

211
00:09:33,340 –> 00:09:37,600
But the more chronic problem
is that any time a peak

212
00:09:37,600 –> 00:09:40,390
approaches capacity a system
becomes mathematically

213
00:09:40,390 –> 00:09:41,380
unstable.

214
00:09:41,380 –> 00:09:43,460
Think of the freeways
at rush hour,

215
00:09:43,460 –> 00:09:45,550
it’s approaching
system capacity,

216
00:09:45,550 –> 00:09:48,302
and therefore, every
metric goes into the tank,

217
00:09:48,302 –> 00:09:50,260
which is what makes it
mathematically unstable.

218
00:09:50,260 –> 00:09:52,420
A 10 minute drive
takes 60 minutes.

219
00:09:52,420 –> 00:09:54,850
A fender bender that should
take 10 minutes to clear

220
00:09:54,850 –> 00:09:56,200
takes two hours to clear.

221
00:09:56,200 –> 00:09:58,158
And a fender bender that
should delay 10 people

222
00:09:58,158 –> 00:09:59,590
delays 10,000 people.

223
00:09:59,590 –> 00:10:02,403
Infusion is a series of fender
benders waiting to happen.

224
00:10:02,403 –> 00:10:04,570
The clinic will run late,
the pharmacy will back up,

225
00:10:04,570 –> 00:10:06,970
the labs will back up, a
nurse will call in sick,

226
00:10:06,970 –> 00:10:09,790
a patient will show up late,
a patient would react badly.

227
00:10:09,790 –> 00:10:12,100
If any of those happen
early on in the day or late

228
00:10:12,100 –> 00:10:13,720
in the afternoon, it’s fine.

229
00:10:13,720 –> 00:10:15,760
If any of them happen in
the middle of the day,

230
00:10:15,760 –> 00:10:19,480
it’s like the big rig crash
at 5:00 PM on a Monday,

231
00:10:19,480 –> 00:10:22,650
the whole system will be
a mess for many hours.

232
00:10:22,650 –> 00:10:25,230
Because of all this,
most infusion centers

233
00:10:25,230 –> 00:10:27,840
have uniquely the
same three problems.

234
00:10:27,840 –> 00:10:29,490
Patients wait a long
time, particularly

235
00:10:29,490 –> 00:10:30,990
in the middle of the day.

236
00:10:30,990 –> 00:10:33,690
The chair profile starts
out narrow, hits a peak,

237
00:10:33,690 –> 00:10:36,990
and comes down, sometimes it
exceeds the chair capacity.

238
00:10:36,990 –> 00:10:39,360
What that means is all the
chairs in an infusion suite

239
00:10:39,360 –> 00:10:41,280
are full, and some chairs
in the waiting room

240
00:10:41,280 –> 00:10:42,690
are occupied as well.

241
00:10:42,690 –> 00:10:45,210
And nurses tend to
either miss their lunch,

242
00:10:45,210 –> 00:10:48,800
or get a granola bar or an apple
and eat lunch at 4 o’clock,

243
00:10:48,800 –> 00:10:51,140
or centers over-staff.

244
00:10:51,140 –> 00:10:53,820
They get float nurses to
cover the lunch break, which

245
00:10:53,820 –> 00:10:56,220
is essentially throwing
nursing labor at the problem,

246
00:10:56,220 –> 00:10:59,710
not really solving the
underlying problem.

247
00:10:59,710 –> 00:11:02,220
So that covers the
first two topics.

248
00:11:02,220 –> 00:11:03,710
In four or five
minutes let me just

249
00:11:03,710 –> 00:11:07,830
cover what an optimized
infusion schedule looks like.

250
00:11:07,830 –> 00:11:10,800
In order to optimize
it, what you have to do

251
00:11:10,800 –> 00:11:14,860
is take this profile
and make a dash profile.

252
00:11:14,860 –> 00:11:17,760
And in order to do that, there’s
a bunch of hard math problems

253
00:11:17,760 –> 00:11:19,140
that need to be solved.

254
00:11:19,140 –> 00:11:21,720
First, with very high
accuracy, predict

255
00:11:21,720 –> 00:11:23,815
the volume for a day of week.

256
00:11:23,815 –> 00:11:24,690
How many on a Monday?

257
00:11:24,690 –> 00:11:26,400
How many on a Tuesday?

258
00:11:26,400 –> 00:11:28,570
Volume in aggregate
is not enough.

259
00:11:28,570 –> 00:11:30,832
You have to predict the
mix very accurately.

260
00:11:30,832 –> 00:11:32,040
How many one hour treatments?

261
00:11:32,040 –> 00:11:33,248
How many two hour treatments?

262
00:11:33,248 –> 00:11:34,560
How many three hour treatments?

263
00:11:34,560 –> 00:11:36,990
Having done that, you
then need to figure out

264
00:11:36,990 –> 00:11:39,060
how much adjustment
you have to make.

265
00:11:39,060 –> 00:11:42,930
So by looking historically
at the last 1,000 times,

266
00:11:42,930 –> 00:11:44,970
you thought something
would be three hours,

267
00:11:44,970 –> 00:11:48,660
and then seeing how accurate you
were, will help you understand

268
00:11:48,660 –> 00:11:50,670
how much you need to adjust it.

269
00:11:50,670 –> 00:11:52,890
And then having
adjusted it, you have

270
00:11:52,890 –> 00:11:56,700
to play the supercomputer
version of Tetris, which

271
00:11:56,700 –> 00:11:59,400
is how do I arrange
all of those blocks

272
00:11:59,400 –> 00:12:03,520
such that I generate the
flattest profile possible?

273
00:12:03,520 –> 00:12:07,650
The reason this becomes
magical is it unlocks capacity.

274
00:12:07,650 –> 00:12:09,990
Unlocking capacity
during peak hours

275
00:12:09,990 –> 00:12:12,720
is like magically finding three
extra lanes on the freeway

276
00:12:12,720 –> 00:12:14,330
that you weren’t expecting.

277
00:12:14,330 –> 00:12:15,760
It gives patient choice.

278
00:12:15,760 –> 00:12:18,360
Meaning, if I needed a
three hour treatment there

279
00:12:18,360 –> 00:12:21,300
are many three hour trains
leaving the station.

280
00:12:21,300 –> 00:12:23,140
It flattens the
workload for nurses

281
00:12:23,140 –> 00:12:26,010
which allows them to be more
productive throughout the day.

282
00:12:26,010 –> 00:12:27,420
And it fits the
nursing schedule.

283
00:12:27,420 –> 00:12:29,100
Nurses who show up
first, leave first.

284
00:12:29,100 –> 00:12:31,050
Nurses who show up
last, leave last.

285
00:12:31,050 –> 00:12:34,400
So this is how the
optimization needs to work.

286
00:12:34,400 –> 00:12:37,730
So what needs to happen,
after all of this,

287
00:12:37,730 –> 00:12:41,600
is to create a template that’s
unique for every center,

288
00:12:41,600 –> 00:12:44,550
for every day, for
every hour of the day.

289
00:12:44,550 –> 00:12:46,940
That’s how precise and
fine-grained the template

290
00:12:46,940 –> 00:12:47,870
needs to be.

291
00:12:47,870 –> 00:12:51,080
And so the way you do that
is think of a simplified

292
00:12:51,080 –> 00:12:54,410
grid where across the top
we’re laying out the duration

293
00:12:54,410 –> 00:12:57,140
buckets, one hour, two
hour, three to five, six

294
00:12:57,140 –> 00:12:58,610
to eight, and nine plus.

295
00:12:58,610 –> 00:13:02,960
We are then saying 10
minute starts, so 7:00

296
00:13:02,960 –> 00:13:06,350
AM, 7:10, these are the
number of patients to start.

297
00:13:06,350 –> 00:13:09,050
The capacity of
the infusion center

298
00:13:09,050 –> 00:13:11,780
limits how many simultaneous
starts it can do.

299
00:13:11,780 –> 00:13:14,450
This center said
between 8:00 and 8:30

300
00:13:14,450 –> 00:13:16,760
they could see three or
four patients at a time.

301
00:13:16,760 –> 00:13:19,680
Once it’s past 9 o’clock they
can only see two at a time.

302
00:13:19,680 –> 00:13:23,360
So no row after 9 o’clock
has more than two starts.

303
00:13:23,360 –> 00:13:26,690
The duration spans
are broad enough

304
00:13:26,690 –> 00:13:28,880
to allow for variability.

305
00:13:28,880 –> 00:13:33,170
And the magic number is
for every duration bucket,

306
00:13:33,170 –> 00:13:36,770
for every start time, how
many patients of that type

307
00:13:36,770 –> 00:13:38,090
should you start.

308
00:13:38,090 –> 00:13:41,570
And so now, the dialogue
with the patient

309
00:13:41,570 –> 00:13:43,120
becomes slightly more nuanced.

310
00:13:43,120 –> 00:13:45,620
Rather than saying, Mrs. Jones,
when would you like to come?

311
00:13:45,620 –> 00:13:47,750
And then just
accepting that time.

312
00:13:47,750 –> 00:13:50,210
The conversation goes, I
see you need a three to five

313
00:13:50,210 –> 00:13:51,260
hour treatment.

314
00:13:51,260 –> 00:13:55,490
I could offer you 8 o’clock,
9:20, 9:40, 10:00, 10:20

315
00:13:55,490 –> 00:13:57,680
or 10:40, would
any of those work?

316
00:13:57,680 –> 00:14:00,320
And so you don’t have to
get 100% of them right,

317
00:14:00,320 –> 00:14:05,510
but if you get 80, 85% of
people into the right buckets,

318
00:14:05,510 –> 00:14:08,110
it’s pre-engineered to
give you a flat profile.

319
00:14:08,110 –> 00:14:09,320
Right?

320
00:14:09,320 –> 00:14:12,440
The reason this
unlocks capacity is

321
00:14:12,440 –> 00:14:14,370
if you look at the before
and after of the way

322
00:14:14,370 –> 00:14:17,930
the Tetris blocks loaded
up, all of this white space

323
00:14:17,930 –> 00:14:21,230
here is lost productivity,
it’s either under-utilization

324
00:14:21,230 –> 00:14:25,010
of a chair, or under-utilization
of a nurse, or both.

325
00:14:25,010 –> 00:14:27,800
On the right hand side, because
you’ve mathematically tightened

326
00:14:27,800 –> 00:14:30,920
the Tetris blocks taking
into account variability,

327
00:14:30,920 –> 00:14:32,620
you unlock productivity.

328
00:14:32,620 –> 00:14:36,650
The productivity can be measured
as the ratio of patient hours

329
00:14:36,650 –> 00:14:37,940
to nursing hours.

330
00:14:37,940 –> 00:14:39,770
So patient hours is
the number of patients

331
00:14:39,770 –> 00:14:41,300
times the average duration.

332
00:14:41,300 –> 00:14:45,600
Nursing hours is the number of
nursing FTE hours on the floor.

333
00:14:45,600 –> 00:14:49,460
If this ratio gets better,
and we can make it better

334
00:14:49,460 –> 00:14:52,910
by 20% or 25%, you
now have choices.

335
00:14:52,910 –> 00:14:56,480
You could choose to monetize
that productivity by absorbing

336
00:14:56,480 –> 00:14:59,120
more volume, which means
you either see more

337
00:14:59,120 –> 00:15:02,840
patients in the day for the
same number of nursing staff,

338
00:15:02,840 –> 00:15:06,560
or you absorb future growth
at either the same staff level

339
00:15:06,560 –> 00:15:08,930
or adding less than you
otherwise might have.

340
00:15:08,930 –> 00:15:11,870
That’s the way you monetize
by absorbing growth,

341
00:15:11,870 –> 00:15:13,890
or you could monetize
by reducing cost.

342
00:15:13,890 –> 00:15:16,430
Meaning, I no longer need to
stay open until 9 o’clock,

343
00:15:16,430 –> 00:15:20,750
I could close at 6:00,
or I could no longer

344
00:15:20,750 –> 00:15:23,570
need to keep a satellite center
open because my main center can

345
00:15:23,570 –> 00:15:24,470
take on more.

346
00:15:24,470 –> 00:15:27,620
So you manage to reduce
cost, or you could just

347
00:15:27,620 –> 00:15:30,560
improve the patient experience,
meaning they wait less

348
00:15:30,560 –> 00:15:33,900
and they get appointments
set at a shorter lead time.

349
00:15:33,900 –> 00:15:36,390
So these are choices
on monetization.

350
00:15:36,390 –> 00:15:38,180
But the key thing
to monetize is it

351
00:15:38,180 –> 00:15:40,010
comes after you create
the productivity

352
00:15:40,010 –> 00:15:41,460
in the first place.

353
00:15:41,460 –> 00:15:41,960
OK.

354
00:15:41,960 –> 00:15:46,880
So that was the quick
refresher of all of the topics

355
00:15:46,880 –> 00:15:49,460
we’ve covered in prior webinars.

356
00:15:49,460 –> 00:15:52,760
So let’s focus now
on the central theme

357
00:15:52,760 –> 00:15:54,530
of today’s webinar,
which is how can you

358
00:15:54,530 –> 00:15:59,100
plan well enough in advance
to run the infusion center.

359
00:15:59,100 –> 00:16:01,170
Infusion center operations
are complicated,

360
00:16:01,170 –> 00:16:03,470
it’s highly variable,
it’s highly dependent,

361
00:16:03,470 –> 00:16:06,180
day for day, hour for
hour, patient for patient.

362
00:16:06,180 –> 00:16:09,450
And so any advanced
intelligence you can get

363
00:16:09,450 –> 00:16:12,550
will let you run the infusion
center much, much better.

364
00:16:12,550 –> 00:16:16,140
So at the very least,
you need to know,

365
00:16:16,140 –> 00:16:20,513
with a fair degree of precision,
how is today going to unfold.

366
00:16:20,513 –> 00:16:21,930
And by a fair
degree of precision,

367
00:16:21,930 –> 00:16:23,710
I mean at 10 minute intervals.

368
00:16:23,710 –> 00:16:26,850
So what you’re seeing here is
if you run the optimization.

369
00:16:26,850 –> 00:16:29,820
The green line represents
what would be optimal.

370
00:16:29,820 –> 00:16:32,040
That’s if you played the
Tetris game right, that’s

371
00:16:32,040 –> 00:16:33,750
the profile you should get.

372
00:16:33,750 –> 00:16:36,360
The bars represent
10 minute windows.

373
00:16:36,360 –> 00:16:38,912
Gray means you’re right
on the optimal frontier.

374
00:16:38,912 –> 00:16:40,870
Yellow means you’re
running a little bit light,

375
00:16:40,870 –> 00:16:42,870
you’re under the
optimal frontier.

376
00:16:42,870 –> 00:16:45,640
If you’d gone above the optimal
it would have been orange.

377
00:16:45,640 –> 00:16:47,460
And if you’d gone
above chair capacity

378
00:16:47,460 –> 00:16:48,432
it would have been red.

379
00:16:48,432 –> 00:16:49,890
That just means
all your chairs are

380
00:16:49,890 –> 00:16:51,750
full for some period of time.

381
00:16:51,750 –> 00:16:54,467
To get this right now,
you can plan your day.

382
00:16:54,467 –> 00:16:56,550
If you’ve got this at 6:00
in the morning, saying,

383
00:16:56,550 –> 00:16:58,800
this is how my day
is going to unfold.

384
00:16:58,800 –> 00:17:03,120
You immediately know that
between 10:30 and 12:20

385
00:17:03,120 –> 00:17:05,849
you’re going to be
running right on par.

386
00:17:05,849 –> 00:17:08,099
It’s probably better not
to do stand-up meetings

387
00:17:08,099 –> 00:17:10,680
or have people take extended
breaks at that time,

388
00:17:10,680 –> 00:17:14,730
kind of run it with an
eye on the operation.

389
00:17:14,730 –> 00:17:17,310
If you needed add-ons,
you know the two windows

390
00:17:17,310 –> 00:17:23,339
between 7:40 and 10 o’clock,
and again between 1:45 and 1:40,

391
00:17:23,339 –> 00:17:27,000
and the rest of the day
you could see your add-ons.

392
00:17:27,000 –> 00:17:29,370
This is how today
is going to unfold.

393
00:17:29,370 –> 00:17:32,070
Equally important is
knowing how the next 30 days

394
00:17:32,070 –> 00:17:33,190
are going to unfold.

395
00:17:33,190 –> 00:17:36,000
So imagine if you could get
the weather map for the next 30

396
00:17:36,000 –> 00:17:39,880
days showing you how
the systems are forming.

397
00:17:39,880 –> 00:17:42,210
So if you can already
tell that 10 days

398
00:17:42,210 –> 00:17:44,450
from now, the middle
of the second row,

399
00:17:44,450 –> 00:17:46,350
I’m going to have a
little bit of heat

400
00:17:46,350 –> 00:17:48,900
where I can see some
orange shaping up.

401
00:17:48,900 –> 00:17:52,110
You can drill down into
that, and start to say,

402
00:17:52,110 –> 00:17:54,330
now I understand
what’s going to happen.

403
00:17:54,330 –> 00:17:56,520
My ramp is slower than
I would have liked,

404
00:17:56,520 –> 00:17:58,650
but it’s going to
lead to congestion

405
00:17:58,650 –> 00:18:01,230
for a 90 minute window
in the middle of the day.

406
00:18:01,230 –> 00:18:04,890
Maybe I should either get a few
nurses to start a bit later,

407
00:18:04,890 –> 00:18:07,540
maybe I can move a
patient to start earlier,

408
00:18:07,540 –> 00:18:09,270
even though that’s
not a preferred thing,

409
00:18:09,270 –> 00:18:12,960
or maybe I can just tell my
schedulers to not schedule

410
00:18:12,960 –> 00:18:15,300
any more because I’m
already running pretty late.

411
00:18:15,300 –> 00:18:19,440
So this is the sort of stuff
that if you had advance warning

412
00:18:19,440 –> 00:18:22,260
you can run the infusion
center much, much better,

413
00:18:22,260 –> 00:18:26,190
versus a reactive mode of
having the staff showing up

414
00:18:26,190 –> 00:18:28,320
in the morning, knowing
that the tsunami is going

415
00:18:28,320 –> 00:18:30,028
to hit sometime in
the middle of the day,

416
00:18:30,028 –> 00:18:32,790
but not knowing when
or why or for how long.

417
00:18:32,790 –> 00:18:35,970
That makes the difference
between a reactive operation

418
00:18:35,970 –> 00:18:39,240
and a proactive
planned operation.

419
00:18:39,240 –> 00:18:42,270
So what does it take to
build a planned operation?

420
00:18:42,270 –> 00:18:45,540
There are two very
different, but very difficult

421
00:18:45,540 –> 00:18:46,860
problems to solve.

422
00:18:46,860 –> 00:18:50,940
The first is a supply
demand matching problem.

423
00:18:50,940 –> 00:18:54,810
What you’ve got to do is,
it’s not sufficient to match

424
00:18:54,810 –> 00:18:56,850
the demand across
the whole day saying,

425
00:18:56,850 –> 00:18:59,130
yep, I can deal with
80 patients in the day.

426
00:18:59,130 –> 00:19:01,020
Well, it matters whether
those 80 patients

427
00:19:01,020 –> 00:19:04,470
came eight per hour for
10 hours, or 20 per hour

428
00:19:04,470 –> 00:19:07,050
for a four hour window
in the middle of the day.

429
00:19:07,050 –> 00:19:10,920
So you’ve got to understand the
arrival pattern, volume, mix,

430
00:19:10,920 –> 00:19:13,650
timing, and try and
match it with your supply

431
00:19:13,650 –> 00:19:15,840
capacity, staff,
equipment, chairs,

432
00:19:15,840 –> 00:19:18,960
et cetera, within a very
tight window, ideally 10 or 15

433
00:19:18,960 –> 00:19:22,350
minutes, maybe up to 30
minutes, despite the fact

434
00:19:22,350 –> 00:19:24,310
that the variability
would be very, very high.

435
00:19:24,310 –> 00:19:26,670
So that’s problem
number one to be solved.

436
00:19:26,670 –> 00:19:29,220
Problem number two to
be solved, is how do you

437
00:19:29,220 –> 00:19:31,110
think about connected services?

438
00:19:31,110 –> 00:19:34,177
When you think about
optimizing connected services,

439
00:19:34,177 –> 00:19:36,510
there are many connected
appointments, labs and clinics,

440
00:19:36,510 –> 00:19:37,770
and procedures.

441
00:19:37,770 –> 00:19:40,830
They have to be
spaced far enough

442
00:19:40,830 –> 00:19:43,500
to be able to execute
it on time every day,

443
00:19:43,500 –> 00:19:45,760
but close enough that it’s
convenient for patients.

444
00:19:45,760 –> 00:19:48,620
If you schedule a connecting
flight through O’Hare every day

445
00:19:48,620 –> 00:19:51,000
with a 15 minute
gap, you know they’ll

446
00:19:51,000 –> 00:19:52,530
miss the connecting flight.

447
00:19:52,530 –> 00:19:54,270
If you schedule it
with a six hour gap,

448
00:19:54,270 –> 00:19:55,980
yes, they’ll make the
connecting flight,

449
00:19:55,980 –> 00:19:59,250
but it’s not efficient or
convenient for the passengers.

450
00:19:59,250 –> 00:20:00,900
So this is the
mathematical balance

451
00:20:00,900 –> 00:20:04,980
you have to strike, close
enough to be executable, but not

452
00:20:04,980 –> 00:20:07,510
so far that it’s inconvenient.

453
00:20:07,510 –> 00:20:09,060
So let’s take each
problem in turn

454
00:20:09,060 –> 00:20:11,310
and describe how it
needs to be solved.

455
00:20:11,310 –> 00:20:15,040
To match supply and demand,
here’s what needs to happen.

456
00:20:15,040 –> 00:20:16,680
You have to analyze
the profiles,

457
00:20:16,680 –> 00:20:20,430
forecast the volume and
mix, and then shape it.

458
00:20:20,430 –> 00:20:23,670
Shape the demand profile by
sequencing the right number

459
00:20:23,670 –> 00:20:25,560
of appointments, off
the right duration,

460
00:20:25,560 –> 00:20:27,240
at each time slot on the day.

461
00:20:27,240 –> 00:20:29,910
That’s kind of what we’ve spent
our time up until now talking

462
00:20:29,910 –> 00:20:33,330
about, which is how do you
build that magic template grid.

463
00:20:33,330 –> 00:20:38,520
The other side of the coin is
sorting out the supply, nurses

464
00:20:38,520 –> 00:20:40,230
and chair availability.

465
00:20:40,230 –> 00:20:43,470
And what you’ve got to do is
figure out the staff ramp-up

466
00:20:43,470 –> 00:20:47,790
schedule, and how you’re going
to allocate chairs, right?

467
00:20:47,790 –> 00:20:50,190
So if you do those
two right, then

468
00:20:50,190 –> 00:20:53,430
what comes out of it
using the mathematics

469
00:20:53,430 –> 00:20:55,150
are two sophisticated models.

470
00:20:55,150 –> 00:20:57,810
One is a forecasting
model on the volume side,

471
00:20:57,810 –> 00:21:00,740
and the second is resource
roles and constraints

472
00:21:00,740 –> 00:21:02,060
on the supply side.

473
00:21:02,060 –> 00:21:04,670
From those, you get
this optimization

474
00:21:04,670 –> 00:21:07,640
to synchronize demand and
supply in very tight windows.

475
00:21:07,640 –> 00:21:11,867
So that’s kind of how you work
the first part of it, right?

476
00:21:11,867 –> 00:21:12,950
Now, what are the supply–

477
00:21:12,950 –> 00:21:14,533
so we’ve talked about
the demand side.

478
00:21:14,533 –> 00:21:16,538
Let’s focus on the supply side.

479
00:21:16,538 –> 00:21:18,830
When you’re going to figure
out nursing capacity, which

480
00:21:18,830 –> 00:21:22,910
is supply constraint 1, the
nurse capacity, you start out

481
00:21:22,910 –> 00:21:27,590
by saying, what is the
typical staffing profile

482
00:21:27,590 –> 00:21:28,760
that you’ve got?

483
00:21:28,760 –> 00:21:29,600
How many shifts?

484
00:21:29,600 –> 00:21:31,280
How many RNs on each shift?

485
00:21:31,280 –> 00:21:34,370
Are the nurses specialized
for specific things,

486
00:21:34,370 –> 00:21:37,610
like port flushes or port draws?

487
00:21:37,610 –> 00:21:38,660
Et cetera.

488
00:21:38,660 –> 00:21:40,890
Do MAs and LPNs
take patients back?

489
00:21:40,890 –> 00:21:42,200
What do charge nurses do?

490
00:21:42,200 –> 00:21:45,590
So you have to understand what
is the resource pool that you

491
00:21:45,590 –> 00:21:47,660
have that you’re working with?

492
00:21:47,660 –> 00:21:52,550
And then second, overlay
[? on ?] your clinical model.

493
00:21:52,550 –> 00:21:56,090
You should not have to change
anyone’s clinical model at all,

494
00:21:56,090 –> 00:21:57,980
because if people say–

495
00:21:57,980 –> 00:22:00,770
if a particular center says, I
like my nurse to be one on one

496
00:22:00,770 –> 00:22:03,553
with the patient for the first
45 minutes of the treatment.

497
00:22:03,553 –> 00:22:04,970
Well, that’s the
way they practice

498
00:22:04,970 –> 00:22:07,170
and they have to be
allowed to do that.

499
00:22:07,170 –> 00:22:09,800
So what you’ve got to do
is say, how many patients

500
00:22:09,800 –> 00:22:11,300
can a single nurse
see at one time

501
00:22:11,300 –> 00:22:13,280
in mid-flight of the treatment?

502
00:22:13,280 –> 00:22:15,530
How much one on one time do
you want at the beginning,

503
00:22:15,530 –> 00:22:17,240
in the middle, and at the end?

504
00:22:17,240 –> 00:22:18,710
How do you think about lunch?

505
00:22:18,710 –> 00:22:21,380
Do you have rules, like no
more than two nurses at a time

506
00:22:21,380 –> 00:22:22,370
taking lunch?

507
00:22:22,370 –> 00:22:24,850
Do you spread the lunch break
out between 11:00 and 2:00?

508
00:22:24,850 –> 00:22:26,390
What exactly do you do?

509
00:22:26,390 –> 00:22:30,770
From all of this, you get
the nursing parameters.

510
00:22:30,770 –> 00:22:33,560
And what we’ve done
on the demand side

511
00:22:33,560 –> 00:22:37,490
is teed it up, so that you’re
starting the right appointments

512
00:22:37,490 –> 00:22:40,490
at the right time to match
this nursing capacity.

513
00:22:40,490 –> 00:22:43,848
You cannot start a patient
unless the nurse is available.

514
00:22:43,848 –> 00:22:45,890
Having a chair available
is insufficient to start

515
00:22:45,890 –> 00:22:46,580
a treatment.

516
00:22:46,580 –> 00:22:48,380
You need both the
chair and the nurse.

517
00:22:48,380 –> 00:22:50,660
And therefore, you
have to orchestrate

518
00:22:50,660 –> 00:22:54,020
how you started such that
both the chair and the nurse

519
00:22:54,020 –> 00:22:55,130
are available.

520
00:22:55,130 –> 00:22:57,620
If you do that right you
will get the flat profile

521
00:22:57,620 –> 00:22:59,090
we are talking about.

522
00:22:59,090 –> 00:23:01,940
And the speed of the
ramp-up is entirely

523
00:23:01,940 –> 00:23:03,900
dependent on your
nurse availability

524
00:23:03,900 –> 00:23:05,240
at the beginning of the day.

525
00:23:05,240 –> 00:23:07,790
During the flat
portion it’s entirely

526
00:23:07,790 –> 00:23:09,980
dependent on how many
simultaneous patients can

527
00:23:09,980 –> 00:23:11,300
a single nurse support.

528
00:23:11,300 –> 00:23:15,210
Those two constraints
determine how high you can go.

529
00:23:15,210 –> 00:23:18,740
So you may have 40 chairs, but
if you have only two nurses,

530
00:23:18,740 –> 00:23:21,140
and say that no nurse can
watch more than four patients

531
00:23:21,140 –> 00:23:23,750
at a time, you will never get–
the flat part of the curve

532
00:23:23,750 –> 00:23:25,470
will never be higher than eight.

533
00:23:25,470 –> 00:23:29,570
So it’s independent of how
many chairs you’ve got.

534
00:23:29,570 –> 00:23:31,280
The chairs are one
constraint, the nurses

535
00:23:31,280 –> 00:23:33,710
are another constraint.

536
00:23:33,710 –> 00:23:35,520
So having done
that, now you’ve got

537
00:23:35,520 –> 00:23:39,140
to think about how nursing
shifts need to be made,

538
00:23:39,140 –> 00:23:42,800
and how do you assign
patients to nurses.

539
00:23:42,800 –> 00:23:45,830
Most conventional
systems end up just

540
00:23:45,830 –> 00:23:48,920
setting up nurse shifts
with simple averaging

541
00:23:48,920 –> 00:23:51,357
and Excel like things,
and say, all right,

542
00:23:51,357 –> 00:23:52,940
let’s have two shifts
one at 8 o’clock

543
00:23:52,940 –> 00:23:54,435
and one starting at 8:30.

544
00:23:54,435 –> 00:23:56,060
Let’s get some number
of nurses at 8:00

545
00:23:56,060 –> 00:23:57,560
and some others at 8:30.

546
00:23:57,560 –> 00:23:59,872
So this is what the
profile looks like.

547
00:23:59,872 –> 00:24:01,580
That looks fine when
you’re just thinking

548
00:24:01,580 –> 00:24:04,520
of it from an HR and a
labor planning perspective.

549
00:24:04,520 –> 00:24:07,820
When you overlay it with
how your demand builds up,

550
00:24:07,820 –> 00:24:10,550
you’ll see you are very
under-utilized in early morning

551
00:24:10,550 –> 00:24:13,490
and you’re potentially
over-utilized at points

552
00:24:13,490 –> 00:24:17,090
in time, where the nurses simply
do not have enough capacity

553
00:24:17,090 –> 00:24:18,770
to deal with their workload.

554
00:24:18,770 –> 00:24:21,650
That creates the rush hour
and domino effect problems,

555
00:24:21,650 –> 00:24:26,300
which then linger on downstream
and for many hours after that.

556
00:24:26,300 –> 00:24:30,530
Many people try and overlay
an acuity model on it.

557
00:24:30,530 –> 00:24:32,850
And acuity models are great.

558
00:24:32,850 –> 00:24:34,970
The problem with
most acuity models

559
00:24:34,970 –> 00:24:36,620
is they tend to
be very aggregate,

560
00:24:36,620 –> 00:24:38,570
and they tend to
be very inexact.

561
00:24:38,570 –> 00:24:41,900
And so what happens is, people
assign an acuity score to one

562
00:24:41,900 –> 00:24:43,640
to five, and then
say, all right,

563
00:24:43,640 –> 00:24:46,820
we want each nurse to have
20 acuity points in the day.

564
00:24:46,820 –> 00:24:49,680
That sounds like a good
and fair thing to do,

565
00:24:49,680 –> 00:24:52,280
but it makes a big difference
whether, as a nurse,

566
00:24:52,280 –> 00:24:55,640
I met my 20 acuity points
[? to ?] 20 patients

567
00:24:55,640 –> 00:24:58,460
of acuity point, one
each, or whether I met it

568
00:24:58,460 –> 00:25:01,460
through four patients of
acuity points, five each.

569
00:25:01,460 –> 00:25:03,470
And so those are
wildly different.

570
00:25:03,470 –> 00:25:05,480
The other part that’s
wildly different,

571
00:25:05,480 –> 00:25:07,310
is it depends on when
my points are getting

572
00:25:07,310 –> 00:25:08,930
accumulated throughout the day.

573
00:25:08,930 –> 00:25:11,180
If all 20 of my
acuity points happened

574
00:25:11,180 –> 00:25:13,940
between 11:00 and 2:00, I had
a very different experience

575
00:25:13,940 –> 00:25:16,250
from a colleague nurse
who’s 20 acuity points are

576
00:25:16,250 –> 00:25:17,990
spread between 8:00 and 6:00.

577
00:25:17,990 –> 00:25:22,880
And so using an aggregate, blunt
instrument, like an acuity,

578
00:25:22,880 –> 00:25:25,940
often feels good, but
doesn’t actually do the job.

579
00:25:25,940 –> 00:25:27,980
If you’re going to
use the acuity points,

580
00:25:27,980 –> 00:25:30,080
they need to be used
in a very precise, very

581
00:25:30,080 –> 00:25:32,430
fine-grained manner.

582
00:25:32,430 –> 00:25:35,570
So what needs to happen is,
when you lay out the loading

583
00:25:35,570 –> 00:25:39,590
profile, you’ve got to try
and set up your shift schedule

584
00:25:39,590 –> 00:25:43,310
to as closely mirror as
possible, both the ramp-up

585
00:25:43,310 –> 00:25:44,850
and the ramp-down phases.

586
00:25:44,850 –> 00:25:48,440
This may require you to set
up more discrete start times,

587
00:25:48,440 –> 00:25:52,010
like nurses who start at 8:00,
8:30, 9:00, 09:30 and 10.

588
00:25:52,010 –> 00:25:54,620
So you may have more
shift starts, which

589
00:25:54,620 –> 00:25:57,780
seems to create complexity
from an HR standpoint,

590
00:25:57,780 –> 00:26:00,530
but it’s important to
balance your workload.

591
00:26:00,530 –> 00:26:04,070
And then, you need
to be able to match

592
00:26:04,070 –> 00:26:08,840
which nurse covers which patient
in a much more precise way.

593
00:26:08,840 –> 00:26:11,942
It’s a holistic workload
that depends on acuity.

594
00:26:11,942 –> 00:26:13,400
It depends on the
number of starts.

595
00:26:13,400 –> 00:26:15,560
It depends on the number
of simultaneous patients.

596
00:26:15,560 –> 00:26:19,160
You cannot use any one of these
metrics to balance out how

597
00:26:19,160 –> 00:26:22,460
the nurses get
assigned to patients,

598
00:26:22,460 –> 00:26:24,800
and how their shifts start.

599
00:26:24,800 –> 00:26:27,083
So that’s how you deal
with supply constraint 1.

600
00:26:27,083 –> 00:26:29,000
If you’re going to deal
with supply constraint

601
00:26:29,000 –> 00:26:33,410
2, which is your
chairs, what’s happening

602
00:26:33,410 –> 00:26:38,210
is the ideal profile is this
trapezoidal shape, where ramps

603
00:26:38,210 –> 00:26:40,770
up stays flat, and ramps down.

604
00:26:40,770 –> 00:26:42,530
You’ll never get a
rectangular shape.

605
00:26:42,530 –> 00:26:44,810
Because in order to get
a rectangular shape,

606
00:26:44,810 –> 00:26:47,040
it would have to be
like a shotgun start,

607
00:26:47,040 –> 00:26:49,953
all 40 infusion chairs are
loaded at 7:00 in the morning.

608
00:26:49,953 –> 00:26:51,620
There’s no chance
that that will happen.

609
00:26:51,620 –> 00:26:53,960
It takes one to two and
two and a half hours

610
00:26:53,960 –> 00:26:55,790
to get the infusion
chairs loaded up.

611
00:26:55,790 –> 00:26:59,040
And so there will always be
a ramp-up and a ramp-down.

612
00:26:59,040 –> 00:27:01,250
The goal is to
reduce leakage, which

613
00:27:01,250 –> 00:27:04,580
means all the white space
between this perfect trapezoid

614
00:27:04,580 –> 00:27:07,100
shape, and the actual
chair utilization,

615
00:27:07,100 –> 00:27:08,370
can potentially be improved.

616
00:27:08,370 –> 00:27:11,150
So that’s kind of what you’re
gunning for, to minimize

617
00:27:11,150 –> 00:27:13,190
this leakage or waste.

618
00:27:13,190 –> 00:27:15,320
The second thing
you can do is if you

619
00:27:15,320 –> 00:27:17,210
want to get more
out of your chairs,

620
00:27:17,210 –> 00:27:19,625
think about how you
speed up your ramp-up.

621
00:27:19,625 –> 00:27:21,500
And there are many things
you can do with it,

622
00:27:21,500 –> 00:27:24,040
one is start earlier, two
is have more nurses on deck

623
00:27:24,040 –> 00:27:24,960
at the beginning.

624
00:27:24,960 –> 00:27:27,320
Most people already have
more nurses than they need.

625
00:27:27,320 –> 00:27:30,390
The challenge is getting enough
patients there in the morning.

626
00:27:30,390 –> 00:27:32,360
And so if there’s
a way to screen out

627
00:27:32,360 –> 00:27:35,060
patients who perhaps don’t need
a clinic appointment in advance

628
00:27:35,060 –> 00:27:37,250
of the infusion,
so the more routine

629
00:27:37,250 –> 00:27:40,592
follow up kinds of infusions,
try and front load those.

630
00:27:40,592 –> 00:27:42,050
So there are various
things you can

631
00:27:42,050 –> 00:27:47,100
do to ramp-up at the front end.

632
00:27:47,100 –> 00:27:48,890
The other thing you
can do is sometimes

633
00:27:48,890 –> 00:27:51,710
infusion centers
ramp-down very gradually

634
00:27:51,710 –> 00:27:53,160
over a period of time.

635
00:27:53,160 –> 00:27:55,710
If you could make the
ramp-down steeper as well,

636
00:27:55,710 –> 00:27:59,510
then you’ve got more area
to fill in the trapezoid.

637
00:27:59,510 –> 00:28:01,790
So think of the more
you color inside

638
00:28:01,790 –> 00:28:03,290
that trapezoid,
the better you are

639
00:28:03,290 –> 00:28:06,710
using your second
constraint of chairs.

640
00:28:06,710 –> 00:28:09,560
And the final thing you
can do, at some point, push

641
00:28:09,560 –> 00:28:11,210
comes to shove,
and you are trying

642
00:28:11,210 –> 00:28:13,130
to put 10 pounds into
a five pound bag,

643
00:28:13,130 –> 00:28:15,420
and there just isn’t
enough capacity.

644
00:28:15,420 –> 00:28:18,870
So once you recognize that,
the first thing you can do,

645
00:28:18,870 –> 00:28:21,835
which is a lower cost is
to just run more hours,

646
00:28:21,835 –> 00:28:23,210
run an extra hour
in the evening,

647
00:28:23,210 –> 00:28:26,100
run two extra hours in the
evening, perhaps run weekends.

648
00:28:26,100 –> 00:28:28,340
So you’re basically
moving the rectangle

649
00:28:28,340 –> 00:28:31,377
to the right, which therefore
makes the trapezoid bigger.

650
00:28:31,377 –> 00:28:32,960
And then finally,
push comes to shove,

651
00:28:32,960 –> 00:28:34,502
you just add more
chairs, which makes

652
00:28:34,502 –> 00:28:36,210
the height of the
rectangle bigger, which

653
00:28:36,210 –> 00:28:37,585
means the height
of the trapezoid

654
00:28:37,585 –> 00:28:38,640
would be bigger as well.

655
00:28:38,640 –> 00:28:42,530
So the trick is to understand
how you can get the most out

656
00:28:42,530 –> 00:28:45,650
of your second supply
constraint, which is chairs.

657
00:28:45,650 –> 00:28:50,630
Now, in order to
analyze it it does

658
00:28:50,630 –> 00:28:54,530
require you to do fairly
sophisticated simulation math.

659
00:28:54,530 –> 00:28:58,160
This is not averaging it out,
and saying I get two point five

660
00:28:58,160 –> 00:29:00,530
patients per chair,
therefore if I add,

661
00:29:00,530 –> 00:29:03,230
you know, two more chairs I
can see five more patients.

662
00:29:03,230 –> 00:29:05,630
That math just breaks
down completely,

663
00:29:05,630 –> 00:29:08,110
because it tends
to be not linear.

664
00:29:08,110 –> 00:29:11,060
It tends to be a
fairly complicated set

665
00:29:11,060 –> 00:29:14,120
of interconnected equations
between capacity, volume,

666
00:29:14,120 –> 00:29:15,090
and supply.

667
00:29:15,090 –> 00:29:18,290
And so what you’ve got to do is
build a simulation model that

668
00:29:18,290 –> 00:29:21,150
says, what if I added
more chairs, what happens?

669
00:29:21,150 –> 00:29:22,380
What if I change my hours?

670
00:29:22,380 –> 00:29:23,900
What if I change my pharmacy?

671
00:29:23,900 –> 00:29:25,490
What if I change my nurse shift?

672
00:29:25,490 –> 00:29:27,818
So each of these what
if questions you’ve

673
00:29:27,818 –> 00:29:29,860
got to be able to run
through a simulation engine

674
00:29:29,860 –> 00:29:31,820
to then come back and
tell you whether it’s

675
00:29:31,820 –> 00:29:34,520
a good idea or a bad idea, or
a high cost idea or a low cost

676
00:29:34,520 –> 00:29:35,420
idea.

677
00:29:35,420 –> 00:29:37,673
So that’s kind of
how you work with it.

678
00:29:37,673 –> 00:29:39,590
So that’s the first part
of the problem, which

679
00:29:39,590 –> 00:29:43,460
is solving the demand
supply matching problem,

680
00:29:43,460 –> 00:29:47,060
treating both nurses and chairs
as a mathematical constraint

681
00:29:47,060 –> 00:29:49,070
into the optimization equation.

682
00:29:49,070 –> 00:29:52,940
The second part is, how do you
optimize connected services?

683
00:29:52,940 –> 00:29:54,470
This is a totally
different problem,

684
00:29:54,470 –> 00:29:56,570
nothing to do with the
demand supply balance.

685
00:29:56,570 –> 00:29:58,070
It’s an equally
complicated problem,

686
00:29:58,070 –> 00:30:00,030
but it’s a completely
different problem.

687
00:30:00,030 –> 00:30:03,720
You have to think of it
as two sub-parts to it.

688
00:30:03,720 –> 00:30:07,277
One is the concept of a
linked connected service.

689
00:30:07,277 –> 00:30:09,110
When you think of the
link connected service

690
00:30:09,110 –> 00:30:11,840
it’s the lab, followed by the
clinic, followed by infusion.

691
00:30:11,840 –> 00:30:13,582
It happens one at
a time, and one

692
00:30:13,582 –> 00:30:15,290
is done before you
can do the second one.

693
00:30:15,290 –> 00:30:17,870
You cannot start the second one
until the first one is done.

694
00:30:17,870 –> 00:30:21,050
This is very similar to
planning the route schedule

695
00:30:21,050 –> 00:30:21,950
for an airline.

696
00:30:21,950 –> 00:30:24,290
So if you are sitting
and thinking through,

697
00:30:24,290 –> 00:30:26,360
what should the 5,000
flights that Delta

698
00:30:26,360 –> 00:30:29,600
does every single day to all
of the cities around the world

699
00:30:29,600 –> 00:30:30,397
need to do?

700
00:30:30,397 –> 00:30:32,480
Then you have to think
about point to point, which

701
00:30:32,480 –> 00:30:33,770
are hubs, which are spokes?

702
00:30:33,770 –> 00:30:35,270
How much connecting
time do you need

703
00:30:35,270 –> 00:30:38,130
to leave in major gateways,
like Atlanta and New York,

704
00:30:38,130 –> 00:30:40,118
and Paris for Delta?

705
00:30:40,118 –> 00:30:42,410
So you have to think through
the entire route schedule,

706
00:30:42,410 –> 00:30:45,530
and therefore the appropriate
layover durations.

707
00:30:45,530 –> 00:30:48,860
The second problem is a
dependent connected service

708
00:30:48,860 –> 00:30:49,550
problem.

709
00:30:49,550 –> 00:30:52,550
What this means is,
all of these pre-things

710
00:30:52,550 –> 00:30:56,520
have to happen before
this service can occur.

711
00:30:56,520 –> 00:30:59,370
So authorization has to happen,
and labs have to happen,

712
00:30:59,370 –> 00:31:02,370
and the pharmacy has to
finish mixing the medications

713
00:31:02,370 –> 00:31:05,010
before you can actually
execute the infusion.

714
00:31:05,010 –> 00:31:08,430
This is very similar to turning
around an aircraft at the gate.

715
00:31:08,430 –> 00:31:10,997
So when a plane pulls in,
lots of things need to happen.

716
00:31:10,997 –> 00:31:13,080
It needs to get refueled,
it needs to get catered,

717
00:31:13,080 –> 00:31:15,670
it need to get cleaned, baggage
handling needs to come in.

718
00:31:15,670 –> 00:31:20,700
And so all of that has to happen
on time in a tight window.

719
00:31:20,700 –> 00:31:23,550
So [? shooting ?] to see
the patient in a chair,

720
00:31:23,550 –> 00:31:26,580
you have to think through, will
all these dependent services

721
00:31:26,580 –> 00:31:29,760
happen on time,
every time, reliably,

722
00:31:29,760 –> 00:31:35,190
so that my chair start
time is as close to reality

723
00:31:35,190 –> 00:31:36,220
as possible?

724
00:31:36,220 –> 00:31:39,540
So these are the two different
network connected problems.

725
00:31:39,540 –> 00:31:43,860
Now, in both of those, you
notice I focused a lot on time.

726
00:31:43,860 –> 00:31:46,860
Meaning, will the linked
appointment happen on time?

727
00:31:46,860 –> 00:31:49,260
Will the dependent
service complete on time?

728
00:31:49,260 –> 00:31:51,720
So why this focus on time?

729
00:31:51,720 –> 00:31:55,890
It turns out, if you go to
optimize any service process,

730
00:31:55,890 –> 00:32:00,360
the intensity of the focus is
entirely on turnaround time.

731
00:32:00,360 –> 00:32:04,500
And here’s why, when you think
about a service, any service,

732
00:32:04,500 –> 00:32:07,350
there are only three
parameters that matter.

733
00:32:07,350 –> 00:32:09,990
How much time did it take
to get that service done?

734
00:32:09,990 –> 00:32:12,180
What did it cost on
a per unit basis?

735
00:32:12,180 –> 00:32:13,557
And what was the error rate?

736
00:32:13,557 –> 00:32:16,140
And deliberately, you’ll see why
I’m calling it error, and not

737
00:32:16,140 –> 00:32:17,190
quality.

738
00:32:17,190 –> 00:32:22,050
So these three corners
capture, in essence,

739
00:32:22,050 –> 00:32:24,060
what the current performance is.

740
00:32:24,060 –> 00:32:26,490
And what you’re
trying to do is push

741
00:32:26,490 –> 00:32:29,940
all three inwards to get to
the best performance possible.

742
00:32:29,940 –> 00:32:32,970
Lower the unit cost of
service, lower the cycle time,

743
00:32:32,970 –> 00:32:34,710
and lower the error
rate, which is why

744
00:32:34,710 –> 00:32:35,877
I wasn’t calling it quality.

745
00:32:35,877 –> 00:32:38,700
I just want to go–
everything towards the inside.

746
00:32:38,700 –> 00:32:43,390
Now, with this is a framework,
let me lay out five assertions.

747
00:32:43,390 –> 00:32:46,285
The first is any
service process,

748
00:32:46,285 –> 00:32:47,910
can be measured on
these three metrics,

749
00:32:47,910 –> 00:32:49,110
time, cost, and error rate.

750
00:32:49,110 –> 00:32:50,370
That’s it.

751
00:32:50,370 –> 00:32:53,170
What you call time, could be
turn around time, total time,

752
00:32:53,170 –> 00:32:54,360
et cetera.

753
00:32:54,360 –> 00:32:56,850
Cost could be fixed cost,
variable cost, total cost,

754
00:32:56,850 –> 00:32:57,790
it doesn’t matter.

755
00:32:57,790 –> 00:33:00,490
These are the only three
dimensions that matter.

756
00:33:00,490 –> 00:33:04,950
The second is the notion
that there is a trade-off

757
00:33:04,950 –> 00:33:06,840
is completely wrong.

758
00:33:06,840 –> 00:33:09,120
Simultaneous improvement
across all three

759
00:33:09,120 –> 00:33:13,450
is not only possible, is
very, very achievable.

760
00:33:13,450 –> 00:33:15,090
The trade-off is
false [? until ?] you

761
00:33:15,090 –> 00:33:16,780
approach perfection.

762
00:33:16,780 –> 00:33:20,640
If you’re executing a service at
a [? Penneys ?] or a [? Pop, ?]

763
00:33:20,640 –> 00:33:25,663
turning around in a millisecond
with a 0.0001 error rate,

764
00:33:25,663 –> 00:33:27,330
it’s possible you’ve
reached the limits,

765
00:33:27,330 –> 00:33:29,350
and you cannot make
it much better.

766
00:33:29,350 –> 00:33:32,280
There are very few services
that achieve that goal,

767
00:33:32,280 –> 00:33:35,640
and therefore, you can improve
all three simultaneously.

768
00:33:35,640 –> 00:33:38,250
Now, if you’re going to improve
all three, which one should

769
00:33:38,250 –> 00:33:40,080
you start with?

770
00:33:40,080 –> 00:33:42,000
Most people start with cost.

771
00:33:42,000 –> 00:33:44,370
But cost is absolutely
the wrong place to start.

772
00:33:44,370 –> 00:33:46,845
Service processes are
labor intensive processes.

773
00:33:46,845 –> 00:33:48,720
If you’re going to try
and take the cost out,

774
00:33:48,720 –> 00:33:50,512
you’re going to have
to take headcount out.

775
00:33:50,512 –> 00:33:53,460
And if you take headcount out,
what happens is you lose skills

776
00:33:53,460 –> 00:33:55,060
and you lose capacity.

777
00:33:55,060 –> 00:33:57,360
And when you lose skills,
more errors happen.

778
00:33:57,360 –> 00:33:59,980
And when you lose capacity,
the time takes longer.

779
00:33:59,980 –> 00:34:03,090
So while you may have succeeded
in pushing in this cost

780
00:34:03,090 –> 00:34:06,270
button of the red buttons,
the other two buttons

781
00:34:06,270 –> 00:34:07,960
have popped out in
the wrong direction.

782
00:34:07,960 –> 00:34:09,418
And so it’s a water
balloon effect,

783
00:34:09,418 –> 00:34:12,080
you push it in one place, it
pushes out in the other two.

784
00:34:12,080 –> 00:34:15,690
Error reduction is also
the wrong place to start.

785
00:34:15,690 –> 00:34:19,560
Because an existing process,
if you overlay error reduction

786
00:34:19,560 –> 00:34:22,170
on it, is only
achieved by inspection.

787
00:34:22,170 –> 00:34:24,155
And when you try and
achieve it by inspection,

788
00:34:24,155 –> 00:34:25,530
you put in checkers,
and then you

789
00:34:25,530 –> 00:34:26,909
put in checkers that
check the checkers, who

790
00:34:26,909 –> 00:34:27,929
check the checkers.

791
00:34:27,929 –> 00:34:29,940
And what happens with
that is both the costs

792
00:34:29,940 –> 00:34:32,538
go up because you’ve got more
checkers involved in the cycle,

793
00:34:32,538 –> 00:34:34,080
and the cycle time
goes up because it

794
00:34:34,080 –> 00:34:36,288
needs to go through more
approval phases for anything

795
00:34:36,288 –> 00:34:36,870
to get done.

796
00:34:36,870 –> 00:34:39,300
So again, it repeats
the water balloon effect

797
00:34:39,300 –> 00:34:41,010
that you pushed in
the error button,

798
00:34:41,010 –> 00:34:43,290
but the other two
buttons popped up.

799
00:34:43,290 –> 00:34:47,100
Time is the only one that
when you push it in it

800
00:34:47,100 –> 00:34:49,050
forces the other two in.

801
00:34:49,050 –> 00:34:51,300
If you suddenly insisted
that the turnaround

802
00:34:51,300 –> 00:34:54,210
time has to be half what
it used to be, guess what?

803
00:34:54,210 –> 00:34:57,600
You no longer have the luxury of
making mistakes and redoing it.

804
00:34:57,600 –> 00:34:59,280
It had better be
right the first time,

805
00:34:59,280 –> 00:35:03,210
and therefore pushes you to
put into place error prevention

806
00:35:03,210 –> 00:35:06,750
mechanisms, and enter detection
mechanisms that minimize it.

807
00:35:06,750 –> 00:35:08,810
If you’re going to say
the cost has to go down–

808
00:35:08,810 –> 00:35:11,790
if you’re going to
say the time goes in,

809
00:35:11,790 –> 00:35:14,430
it forces you to think about
a workflow that minimizes

810
00:35:14,430 –> 00:35:16,770
hand-offs, because
hand-offs mean more people,

811
00:35:16,770 –> 00:35:18,440
more people means more cost.

812
00:35:18,440 –> 00:35:21,278
So time is the only one
when you push the button in,

813
00:35:21,278 –> 00:35:23,070
instead of repelling
the other two buttons,

814
00:35:23,070 –> 00:35:24,720
it pulls the other
two buttons in.

815
00:35:24,720 –> 00:35:27,250
And that’s why it’s the
right place to start.

816
00:35:27,250 –> 00:35:31,230
Now, let me make this come
to life with an analogy.

817
00:35:31,230 –> 00:35:33,450
When you think about
manufacturing– what happens

818
00:35:33,450 –> 00:35:37,650
is, oftentimes the problem is
more invisible than visible.

819
00:35:37,650 –> 00:35:40,060
So think about manufacturing.

820
00:35:40,060 –> 00:35:44,640
If I had a lot of
inventory, then I

821
00:35:44,640 –> 00:35:47,460
could have a lot of
problems in my process.

822
00:35:47,460 –> 00:35:49,950
I could have bad suppliers, I
could have production quality

823
00:35:49,950 –> 00:35:52,470
problems, my machines could
break down all the time.

824
00:35:52,470 –> 00:35:55,710
But anytime I got an order
to supply 100 widgets,

825
00:35:55,710 –> 00:35:57,810
I had a million widgets
sitting in my warehouse

826
00:35:57,810 –> 00:35:59,850
and I could ship
from my inventory.

827
00:35:59,850 –> 00:36:02,040
And so what the
inventory did, just

828
00:36:02,040 –> 00:36:04,440
like the tip of the
iceberg, it allowed

829
00:36:04,440 –> 00:36:07,710
me to get very sloppy on a
bunch of the other metrics

830
00:36:07,710 –> 00:36:11,070
and seemingly appear to
meet performance objectives

831
00:36:11,070 –> 00:36:12,810
even though my
underlying manufacturing

832
00:36:12,810 –> 00:36:14,320
process was broken.

833
00:36:14,320 –> 00:36:17,100
If you took away my
inventory, then suddenly when

834
00:36:17,100 –> 00:36:19,020
I get an order for
100 widgets, I’m

835
00:36:19,020 –> 00:36:21,540
not able to produce it
because my suppliers are bad,

836
00:36:21,540 –> 00:36:23,910
my compliance is horrible,
my production is terrible,

837
00:36:23,910 –> 00:36:29,670
and my equipment
utilization is bad.

838
00:36:29,670 –> 00:36:34,810
Similarly, in a service process,
cycle time is what drives it.

839
00:36:34,810 –> 00:36:37,890
So if I have a
long cycle time, I

840
00:36:37,890 –> 00:36:40,370
can have all kinds
of problems below it.

841
00:36:40,370 –> 00:36:44,050
So imagine if I took 30
days to process a bank loan.

842
00:36:44,050 –> 00:36:46,140
I could get my forms
wrong five times.

843
00:36:46,140 –> 00:36:47,520
I could get my
inspections wrong.

844
00:36:47,520 –> 00:36:49,560
I could get the
application wrong.

845
00:36:49,560 –> 00:36:51,300
Because the amount
of work required

846
00:36:51,300 –> 00:36:54,570
is only two hours of work, and
I had 30 days to deal with it.

847
00:36:54,570 –> 00:36:56,220
So when you force
the time to zero,

848
00:36:56,220 –> 00:36:58,810
you force a superior
performance.

849
00:36:58,810 –> 00:37:03,300
So that’s why in both connected
services and dependent services

850
00:37:03,300 –> 00:37:05,040
you should focus on cycle time.

851
00:37:05,040 –> 00:37:06,600
Are you meeting
the turnaround time

852
00:37:06,600 –> 00:37:08,160
expected for labs or pharmacy?

853
00:37:08,160 –> 00:37:09,060
Et cetera.

854
00:37:09,060 –> 00:37:11,440
Let me give you an example.

855
00:37:11,440 –> 00:37:15,550
For labs, when you try and say I
want to minimize the turnaround

856
00:37:15,550 –> 00:37:19,030
time off of a lab
blood drop, then

857
00:37:19,030 –> 00:37:21,100
the three variables
that are involved

858
00:37:21,100 –> 00:37:25,000
is, at what rates are our
patients arriving into the lab?

859
00:37:25,000 –> 00:37:27,130
How many staff do you
have to deal with it?

860
00:37:27,130 –> 00:37:29,380
Meaning, phlebotomists
and nurses et cetera.

861
00:37:29,380 –> 00:37:32,330
And how quickly are you
processing each patient?

862
00:37:32,330 –> 00:37:34,600
These are the only three
variables that matter.

863
00:37:34,600 –> 00:37:37,150
And there are very tight
mathematical equations

864
00:37:37,150 –> 00:37:38,850
that connected, right?

865
00:37:38,850 –> 00:37:40,660
They’re messy, nasty
equations, but they’re

866
00:37:40,660 –> 00:37:42,890
very well-defined equations.

867
00:37:42,890 –> 00:37:46,210
And when you do this, it lets
you balance supply and demand

868
00:37:46,210 –> 00:37:48,790
in a 15 minute window,
which ties back

869
00:37:48,790 –> 00:37:52,752
how connected services play
back into the demand and supply

870
00:37:52,752 –> 00:37:54,460
balance, which is why
you needed to solve

871
00:37:54,460 –> 00:37:57,080
both of those problems.

872
00:37:57,080 –> 00:38:00,630
At a large academic medical
center, we did this for labs.

873
00:38:00,630 –> 00:38:03,732
Their issue was the labs were
taking an hour to process,

874
00:38:03,732 –> 00:38:05,940
and it was regardless of
whether they were getting it

875
00:38:05,940 –> 00:38:10,080
from the needle
or from the port.

876
00:38:10,080 –> 00:38:14,670
And so we first analyzed the
arrival patterns by hour of day

877
00:38:14,670 –> 00:38:16,330
at 15 minute intervals.

878
00:38:16,330 –> 00:38:18,150
So we knew how many
port draw patients

879
00:38:18,150 –> 00:38:20,847
versus regular phlebotomist
patients are coming in,

880
00:38:20,847 –> 00:38:22,680
took a whole bunch of
actions, some of which

881
00:38:22,680 –> 00:38:25,530
they had done better.

882
00:38:25,530 –> 00:38:28,440
And having done it
better, we were then

883
00:38:28,440 –> 00:38:32,970
able to give them an optimized
schedule of when nurses should

884
00:38:32,970 –> 00:38:35,160
start, when phlebotomists
should start,

885
00:38:35,160 –> 00:38:38,520
give them the training to
minimize the turnaround

886
00:38:38,520 –> 00:38:40,350
time for processing an order.

887
00:38:40,350 –> 00:38:42,810
And as a result of all
of those actions, here’s

888
00:38:42,810 –> 00:38:44,910
what happened over a fairly
short period of eight

889
00:38:44,910 –> 00:38:48,240
to 12 weeks, the turnaround
time for lab draw

890
00:38:48,240 –> 00:38:52,620
went from 50 to 60 minutes,
down to 15 to 20 minutes, so 60

891
00:38:52,620 –> 00:38:55,500
plus percent reduction
in time, with no drop

892
00:38:55,500 –> 00:38:58,600
in quality or cost.

893
00:38:58,600 –> 00:39:01,200
In fact, improvements
on both of those.

894
00:39:01,200 –> 00:39:05,010
So that’s kind of how all of
these pieces fit together.

895
00:39:05,010 –> 00:39:08,940
The webinar coming up
in a couple of weeks,

896
00:39:08,940 –> 00:39:09,930
is on diagnostics.

897
00:39:09,930 –> 00:39:11,730
Let me give you a
two minute preview

898
00:39:11,730 –> 00:39:15,060
of what you can expect to see.

899
00:39:15,060 –> 00:39:16,860
In order to analyze
the performance,

900
00:39:16,860 –> 00:39:19,110
there’s a whole bunch of
intelligent diagnostics

901
00:39:19,110 –> 00:39:23,580
that need to be done, that start
with understanding the volume

902
00:39:23,580 –> 00:39:25,830
pattern, understanding the mix.

903
00:39:25,830 –> 00:39:28,170
How often is the
volume pattern getting

904
00:39:28,170 –> 00:39:29,610
close to the edge of the cliff?

905
00:39:29,610 –> 00:39:31,860
Because when you get close
to the edge of the cliff,

906
00:39:31,860 –> 00:39:36,630
is when the wait
times go up and so on.

907
00:39:36,630 –> 00:39:39,510
How much is the add on
and cancellation rate?

908
00:39:39,510 –> 00:39:42,030
Which means the inherent
variability and instability

909
00:39:42,030 –> 00:39:43,260
with schedule.

910
00:39:43,260 –> 00:39:46,650
Is it impacting you adversely
or is it a modest enough shock

911
00:39:46,650 –> 00:39:48,270
to the system that
you can absorb?

912
00:39:48,270 –> 00:39:52,592
You need to understand that
by being able to look at it.

913
00:39:52,592 –> 00:39:54,300
You need to be able
to tell, are patients

914
00:39:54,300 –> 00:39:57,420
arriving early or late, and
what’s happening with that.

915
00:39:57,420 –> 00:40:01,410
And finally you need to tell
whether the guidance on how

916
00:40:01,410 –> 00:40:06,900
to steer scheduling is being,
A, followed, and B, executed.

917
00:40:06,900 –> 00:40:09,990
And if it’s being followed and
the results are still not good,

918
00:40:09,990 –> 00:40:12,000
then maybe the
template is wrong.

919
00:40:12,000 –> 00:40:14,700
If it’s not being followed,
and the results aren’t good,

920
00:40:14,700 –> 00:40:16,420
then maybe you should
follow it first.

921
00:40:16,420 –> 00:40:17,850
And once we follow
it we can start

922
00:40:17,850 –> 00:40:19,720
to see if the template
is good or not.

923
00:40:19,720 –> 00:40:23,710
And so you need to build a
closed loop learning system,

924
00:40:23,710 –> 00:40:26,130
which starts with the
optimization puts out

925
00:40:26,130 –> 00:40:28,350
a guidance for how
you should schedule,

926
00:40:28,350 –> 00:40:31,200
monitors how you’re actually
scheduling against that,

927
00:40:31,200 –> 00:40:33,720
connects it to the performance,
and then learns from it

928
00:40:33,720 –> 00:40:35,160
and tweaks the template.

929
00:40:35,160 –> 00:40:36,990
When you build
this learning loop,

930
00:40:36,990 –> 00:40:38,700
it’s a bit like
a thermostat that

931
00:40:38,700 –> 00:40:41,100
knows when the temperature’s
above what you set it,

932
00:40:41,100 –> 00:40:43,050
as it needs to fire up
the air conditioning.

933
00:40:43,050 –> 00:40:44,770
And when the temperature’s
below what you set it as,

934
00:40:44,770 –> 00:40:46,395
it needs to fire up
the heating system.

935
00:40:46,395 –> 00:40:48,780
And it constantly alternates
between the heating

936
00:40:48,780 –> 00:40:51,720
and cooling, until you get
it stable at the desired

937
00:40:51,720 –> 00:40:52,610
temperature.

938
00:40:52,610 –> 00:40:54,110
The desired temperature
for what you

939
00:40:54,110 –> 00:40:56,130
are shooting for
an infusion center,

940
00:40:56,130 –> 00:40:59,070
is that perfect trapezoid
of a smooth ramp-up,

941
00:40:59,070 –> 00:41:04,070
flat throughout the day,
and a smooth ramp-down.

942
00:41:04,070 –> 00:41:07,130
So that’s the content for today.

943
00:41:07,130 –> 00:41:09,230
In the last five
or 10 minutes, let

944
00:41:09,230 –> 00:41:14,890
me focus on incoming questions.

945
00:41:14,890 –> 00:41:20,310
The first question is, are
the previous webinar slides

946
00:41:20,310 –> 00:41:20,920
available?

947
00:41:20,920 –> 00:41:25,830
Yes, when you– the same link
that you used to register

948
00:41:25,830 –> 00:41:29,430
has the previous webinars
available for downloading.

949
00:41:29,430 –> 00:41:33,300
And so webinars 1 and 2 are
available for downloading,

950
00:41:33,300 –> 00:41:35,490
which would give you the
more detailed views of what

951
00:41:35,490 –> 00:41:38,310
I rushed through in the
first five or 10 minutes.

952
00:41:38,310 –> 00:41:41,190
And after today, by sometime
tomorrow, today’s webinar

953
00:41:41,190 –> 00:41:42,190
will be loaded as well.

954
00:41:42,190 –> 00:41:45,570
So within 24 hours after
the end of the webinar

955
00:41:45,570 –> 00:41:47,260
it gets loaded up.

956
00:41:47,260 –> 00:41:48,820
OK.

957
00:41:48,820 –> 00:41:50,160
Question.

958
00:41:50,160 –> 00:41:53,970
Our nurses are
unionized, is it still

959
00:41:53,970 –> 00:41:56,310
possible to optimize schedules?

960
00:41:56,310 –> 00:41:58,570
Yes, very much so.

961
00:41:58,570 –> 00:42:01,140
Even in a unionized
environment you

962
00:42:01,140 –> 00:42:05,080
can set shift
schedules and so on.

963
00:42:05,080 –> 00:42:07,860
Now, you can’t change them
day to day and week to week,

964
00:42:07,860 –> 00:42:09,720
you have to plan
it well in advance

965
00:42:09,720 –> 00:42:11,740
and get approvals and so on.

966
00:42:11,740 –> 00:42:13,390
So you can do that.

967
00:42:13,390 –> 00:42:18,030
And so what will happen
is, if you build up

968
00:42:18,030 –> 00:42:20,050
a system of just 10
and 12 hour shifts

969
00:42:20,050 –> 00:42:21,990
then you’d want a bunch
of eight hour shifts,

970
00:42:21,990 –> 00:42:24,365
then that might take you a
little bit longer to negotiate

971
00:42:24,365 –> 00:42:27,450
with the union to get the right
composition of nursing shifts.

972
00:42:27,450 –> 00:42:31,680
But as long as you offer up
enough lead time, and give

973
00:42:31,680 –> 00:42:36,090
people the choice to work a
shift that they want to work,

974
00:42:36,090 –> 00:42:41,790
the unionized environment does
not necessarily limit to you.

975
00:42:41,790 –> 00:42:45,960
New question, what are the
challenges around the add-ons

976
00:42:45,960 –> 00:42:47,070
to the current schedule?

977
00:42:47,070 –> 00:42:52,530
Yes, infusion is a fairly
high add-on, and not so

978
00:42:52,530 –> 00:42:55,770
high, but suddenly occurs,
no-show and cancellation rate.

979
00:42:55,770 –> 00:43:00,160
We typically see no-show and
cancellation rates of 6% or 7%,

980
00:43:00,160 –> 00:43:03,910
and add-on rates roughly that or
sometimes a little bit higher.

981
00:43:03,910 –> 00:43:06,750
Now, what needs to
happen is neither

982
00:43:06,750 –> 00:43:10,980
an add-on, nor a cancellation,
is precisely predictable.

983
00:43:10,980 –> 00:43:13,502
You can predict that x
number are likely to,

984
00:43:13,502 –> 00:43:15,210
but that’s not helpful,
because you don’t

985
00:43:15,210 –> 00:43:16,780
know when it’s going to happen.

986
00:43:16,780 –> 00:43:20,190
So what you need to do, is
for each day of the week,

987
00:43:20,190 –> 00:43:22,290
you have to get a
very good sense of,

988
00:43:22,290 –> 00:43:26,730
is my cancellation envelope
higher than my add-on envelope?

989
00:43:26,730 –> 00:43:29,100
Meaning, more people are
likely to cancel or no-show

990
00:43:29,100 –> 00:43:30,540
than are likely to be added on.

991
00:43:30,540 –> 00:43:32,850
If so, you’ve got
an opportunity.

992
00:43:32,850 –> 00:43:35,110
Add-ons should not be
dealt with on a first come,

993
00:43:35,110 –> 00:43:36,270
first serve basis.

994
00:43:36,270 –> 00:43:38,730
Add-ons you need to be
very strategic about.

995
00:43:38,730 –> 00:43:41,190
If you’ve got to the
[? huddled ?] profile that

996
00:43:41,190 –> 00:43:43,560
shows when you’ve
got pockets of time,

997
00:43:43,560 –> 00:43:45,300
then when you get
an add-on request,

998
00:43:45,300 –> 00:43:47,880
you can confidently steer it to
one of those windows of time,

999
00:43:47,880 –> 00:43:49,530
knowing you’ll be fine.

1000
00:43:49,530 –> 00:43:54,060
If you’re trying to steer it
into a window that’s crowded,

1001
00:43:54,060 –> 00:43:56,670
rather than just taking it
on, crossing your fingers,

1002
00:43:56,670 –> 00:43:59,253
and hoping for the best, because
that’s what happens every day

1003
00:43:59,253 –> 00:44:01,680
and it never works out
right, is to create some kind

1004
00:44:01,680 –> 00:44:03,155
of a dynamic wait list.

1005
00:44:03,155 –> 00:44:04,530
Where you tell
the person, Yes, I

1006
00:44:04,530 –> 00:44:06,947
know you need to be added on
sometime in the 12:00 to 3:00

1007
00:44:06,947 –> 00:44:07,530
window.

1008
00:44:07,530 –> 00:44:09,390
I don’t have an
exact slot for you

1009
00:44:09,390 –> 00:44:12,300
yet, give me your cell
phone, I’ll text you.

1010
00:44:12,300 –> 00:44:14,820
Because you know your
cancellations will occur,

1011
00:44:14,820 –> 00:44:17,490
there will be a cancellation
or a no-show sometime

1012
00:44:17,490 –> 00:44:20,970
in that window that happens
to match the duration.

1013
00:44:20,970 –> 00:44:23,970
The biggest mistake you
can make is jam an add-on

1014
00:44:23,970 –> 00:44:25,710
into an available chair.

1015
00:44:25,710 –> 00:44:28,157
The add-on could be
a five hour opening,

1016
00:44:28,157 –> 00:44:29,990
and the chair opening
based on optimization,

1017
00:44:29,990 –> 00:44:31,082
is a two hour opening.

1018
00:44:31,082 –> 00:44:32,790
You’ve now jammed a
five hour appointment

1019
00:44:32,790 –> 00:44:35,340
into a two hour slot that
will create a domino effect

1020
00:44:35,340 –> 00:44:37,080
downstream and
mess everything up.

1021
00:44:37,080 –> 00:44:40,320
So you’re much better off
waiting for a five hour

1022
00:44:40,320 –> 00:44:41,850
cancel to occur,
because you know

1023
00:44:41,850 –> 00:44:45,330
with confidence that a five
or six hour cancel will likely

1024
00:44:45,330 –> 00:44:46,050
occur.

1025
00:44:46,050 –> 00:44:47,400
And then you steer it.

1026
00:44:47,400 –> 00:44:50,730
In some ways this is exactly
how restaurants deal with it.

1027
00:44:50,730 –> 00:44:53,280
If you go to a restaurant
and ask for a table for two,

1028
00:44:53,280 –> 00:44:56,190
and then behind you comes
somebody as a group of six.

1029
00:44:56,190 –> 00:44:58,530
And a table for six
opens up, guess what?

1030
00:44:58,530 –> 00:45:01,080
The restaurant doesn’t put you
in it because you came first,

1031
00:45:01,080 –> 00:45:03,930
it puts the party of six
at the table for six.

1032
00:45:03,930 –> 00:45:06,330
And even though you realize
they came behind you,

1033
00:45:06,330 –> 00:45:08,340
nobody complains about
it because you say, yep,

1034
00:45:08,340 –> 00:45:10,890
that’s a better utilization
of a table for six

1035
00:45:10,890 –> 00:45:11,790
at the restaurant.

1036
00:45:11,790 –> 00:45:16,440
That’s the kind of mindset
that needs to be applied here.

1037
00:45:16,440 –> 00:45:19,670
Question number
four, how do we know

1038
00:45:19,670 –> 00:45:24,010
when it’s time to add chairs?

1039
00:45:24,010 –> 00:45:25,130
Right.

1040
00:45:25,130 –> 00:45:27,800
So how do you know when
it’s time to add chairs?

1041
00:45:27,800 –> 00:45:31,460
So if you recall I walked
through the four levers

1042
00:45:31,460 –> 00:45:34,730
you pull to get the most
out of the existing chairs,

1043
00:45:34,730 –> 00:45:37,250
one is fill the trapezoid
as best as you can,

1044
00:45:37,250 –> 00:45:39,780
two is ramp-up
faster if you can,

1045
00:45:39,780 –> 00:45:42,910
three is ramp-down more abruptly
towards the end of the day

1046
00:45:42,910 –> 00:45:46,790
if you can, and four is add some
hours, then add some chairs.

1047
00:45:46,790 –> 00:45:49,880
At some point, yes, everybody
does need to add chairs.

1048
00:45:49,880 –> 00:45:52,400
And you need to plan it
with enough lead time,

1049
00:45:52,400 –> 00:45:54,980
knowing that it
takes time to get

1050
00:45:54,980 –> 00:45:58,382
expansion built and commissioned
and operational et cetera.

1051
00:45:58,382 –> 00:45:59,840
So if you assumed
it would take you

1052
00:45:59,840 –> 00:46:01,400
six months to get
that going, you

1053
00:46:01,400 –> 00:46:04,430
need to be forecasting
volume, and what you

1054
00:46:04,430 –> 00:46:06,620
can do well enough
in advance that it

1055
00:46:06,620 –> 00:46:10,410
happens without a hiccup.

1056
00:46:10,410 –> 00:46:15,150
Does your system communicate
with any of the well-known HRs?

1057
00:46:15,150 –> 00:46:17,850
We are completely
agnostic to the HR.

1058
00:46:17,850 –> 00:46:20,320
So our deployments span
[? Epics, ?] [? Turner, ?]

1059
00:46:20,320 –> 00:46:24,250
[INAUDIBLE],, Mosaic,
[INAUDIBLE],, ARIA, all of them.

1060
00:46:24,250 –> 00:46:26,820
And so the reason we
do that, is we don’t

1061
00:46:26,820 –> 00:46:28,350
need to integrate with any HR.

1062
00:46:28,350 –> 00:46:32,020
We build it so that we are not
dependent on the HR at all.

1063
00:46:32,020 –> 00:46:35,340
What our customers do
is they pull the data

1064
00:46:35,340 –> 00:46:37,620
through standard chair
reports from their existing

1065
00:46:37,620 –> 00:46:41,610
HR, whatever it is, package
it like a CSV file or an Excel

1066
00:46:41,610 –> 00:46:43,950
spreadsheet, and push it to us.

1067
00:46:43,950 –> 00:46:47,530
All the optimization and
algorithms take place with us.

1068
00:46:47,530 –> 00:46:50,520
And then we don’t want
schedulers using our system.

1069
00:46:50,520 –> 00:46:53,670
Schedulers are very well
trained on using HR scheduling

1070
00:46:53,670 –> 00:46:55,620
system and all the
other information

1071
00:46:55,620 –> 00:46:58,330
they need about the clinical
realities of that patient

1072
00:46:58,330 –> 00:46:59,310
are on the HR.

1073
00:46:59,310 –> 00:47:01,860
So what we do is we
take our recommendation

1074
00:47:01,860 –> 00:47:06,270
and put it as templates into
the EHR’s current templating

1075
00:47:06,270 –> 00:47:07,320
mechanism.

1076
00:47:07,320 –> 00:47:11,550
Today’s templates in all the
customers– on all infusion

1077
00:47:11,550 –> 00:47:13,307
centers that are
not our customers,

1078
00:47:13,307 –> 00:47:14,890
have been built by
inspection, rights?

1079
00:47:14,890 –> 00:47:17,490
So somebody somewhere
decided that at 8 o’clock

1080
00:47:17,490 –> 00:47:20,310
we’d start four infusions, at
8:30 we would start four more,

1081
00:47:20,310 –> 00:47:22,110
at 9 o’clock we’d
start six more.

1082
00:47:22,110 –> 00:47:23,830
And so that template
has been built

1083
00:47:23,830 –> 00:47:27,660
with just this kind of simple
minded mathematical inspection.

1084
00:47:27,660 –> 00:47:32,160
We’re replacing that logic
with the deep optimization

1085
00:47:32,160 –> 00:47:35,280
logic that starts
to say, at 7:10 you

1086
00:47:35,280 –> 00:47:37,828
should start one 1 hour
appointment and two 3 hour

1087
00:47:37,828 –> 00:47:38,370
appointments.

1088
00:47:38,370 –> 00:47:41,650
At 7:20 you should do one 5
hour appointment, and so on.

1089
00:47:41,650 –> 00:47:45,150
And so we replace the simple
math with sophisticated math,

1090
00:47:45,150 –> 00:47:48,223
but otherwise co-exist
with any HR, regardless.

1091
00:47:51,930 –> 00:47:52,802
Pause for a moment.

1092
00:47:52,802 –> 00:47:54,010
Any more questions coming up?

1093
00:48:02,500 –> 00:48:03,950
Not seeing any come across.

1094
00:48:06,950 –> 00:48:09,300
MODERATOR: No I think that’s it.

1095
00:48:09,300 –> 00:48:11,930
If you have any
questions post webinar,

1096
00:48:11,930 –> 00:48:13,550
you can still type
them in and we

1097
00:48:13,550 –> 00:48:15,750
will receive them post webinar.

1098
00:48:15,750 –> 00:48:19,520
So if anything pops into your
head, please let us know.

1099
00:48:19,520 –> 00:48:23,420
And again, you can contact us
directly using the text number

1100
00:48:23,420 –> 00:48:25,460
at the top of your
console, and also the email

1101
00:48:25,460 –> 00:48:27,720
address that’s at the
top of your console.

1102
00:48:27,720 –> 00:48:32,570
So special thanks to Mohan for
presenting today’s webinar.

1103
00:48:32,570 –> 00:48:35,900
Friendly reminder, a recording
of this webinar with the slides

1104
00:48:35,900 –> 00:48:39,230
should appear in your
inbox in about 24 hours.

1105
00:48:39,230 –> 00:48:42,500
And thanks again to all of
you for joining us today.