WEBVTT

88f70259-9efd-438f-93a0-1d752834c1ad-0
00:00:00.600 --> 00:00:05.997
This is the first occupation where I've
never heard anybody give me a hard time

88f70259-9efd-438f-93a0-1d752834c1ad-1
00:00:05.997 --> 00:00:10.180
for second guessing something,
for asking too many questions,

88f70259-9efd-438f-93a0-1d752834c1ad-2
00:00:10.180 --> 00:00:14.769
for not being satisfied with the first
answer that we come up with,

88f70259-9efd-438f-93a0-1d752834c1ad-3
00:00:14.769 --> 00:00:18.480
or for coming up with a new question or a
new problem.

79a05303-5d46-4b0b-9318-7d8325c7e7e7-0
00:00:19.000 --> 00:00:21.040
I mean, I've I found my calling.

6c583034-dc48-4075-82d9-7869b7ef4839-0
00:00:23.960 --> 00:00:28.440
My name is Katie Eckstrom Gran,
and I'm a communication pathologist.

41ad214f-16df-455a-8bdb-80b0a46aaec6-0
00:00:28.720 --> 00:00:30.560
I mean, I could go on and on about this.

3bb7fb0e-429c-4321-890b-727368addef9-0
00:00:30.840 --> 00:00:32.040
I don't even know where to start.

c0f9d8aa-c029-4158-b09c-1c97d0844a17-0
00:00:32.600 --> 00:00:33.680
I'm going to start in the middle.

791ba92d-f96a-4627-a7bb-e05c1ecb98d1-0
00:00:35.720 --> 00:00:39.737
I really lucked out in so many ways
because I'm in this interprofessional

791ba92d-f96a-4627-a7bb-e05c1ecb98d1-1
00:00:39.737 --> 00:00:41.040
Health Sciences program.

c83234d0-d51d-416d-8694-4d4d3d745633-0
00:00:41.280 --> 00:00:47.223
My research is to be able to develop ways
to describe the most important

c83234d0-d51d-416d-8694-4d4d3d745633-1
00:00:47.223 --> 00:00:53.330
characteristics of natural communication
in ways that we can then apply to

c83234d0-d51d-416d-8694-4d4d3d745633-2
00:00:53.330 --> 00:00:57.320
supporting improving communication for
everyone.

dc1db6b1-bf25-4348-aa03-987ed5a2eba5-0
00:00:57.840 --> 00:01:03.477
My ultimate goal is to be able to apply
what we find to improving people's

dc1db6b1-bf25-4348-aa03-987ed5a2eba5-1
00:01:03.477 --> 00:01:04.680
quality of life.

6f50196e-dd38-4200-8ef8-f3709551d815-0
00:01:06.240 --> 00:01:09.115
I mean,
I wish I had started sooner because this

6f50196e-dd38-4200-8ef8-f3709551d815-1
00:01:09.115 --> 00:01:12.520
is so fun and there's so much to do,
but I couldn't have.

0ef3ce23-f7b9-4893-8133-068372443ff7-0
00:01:13.200 --> 00:01:18.530
I think we needed this technological
computational revolution to ask kinds of

0ef3ce23-f7b9-4893-8133-068372443ff7-1
00:01:18.530 --> 00:01:23.040
questions that are really important for
improving human outcomes.

77695332-a313-4d0a-a954-a03144d56cf6-0
00:01:24.400 --> 00:01:30.453
But speech pathologists aren't trained to
do these kind of large scale novel

77695332-a313-4d0a-a954-a03144d56cf6-1
00:01:30.453 --> 00:01:35.800
computational strategies,
both in the kind of initial data analysis

77695332-a313-4d0a-a954-a03144d56cf6-2
00:01:35.800 --> 00:01:40.360
and then the statistical analysis in the
Glass brain lab.

158b5e5e-70f6-477d-bc31-e5228fcca4d4-0
00:01:40.680 --> 00:01:44.684
I mean,
they're engineers that can solve data

158b5e5e-70f6-477d-bc31-e5228fcca4d4-1
00:01:44.684 --> 00:01:51.040
problems or data questions or data
conflicts and make new things, right?

9e0fb3b9-4220-42f6-a883-9e6e674f9b6e-0
00:01:51.040 --> 00:01:53.800
So we've made, they call it a pipeline,
right?

c3b7b713-e2de-4619-95a9-f41ee81c0aa0-0
00:01:53.800 --> 00:01:58.443
So we've made a whole new way to process
files, audio files,

c3b7b713-e2de-4619-95a9-f41ee81c0aa0-1
00:01:58.443 --> 00:02:03.240
going all the way from I'm having a
conversation with someone.

6ca4c6c9-9da1-4126-b4bf-053386ee7141-0
00:02:03.480 --> 00:02:10.120
And then how do we get from there to all
these different streams of usable data?

c70532fc-751c-4a11-b841-26a5cff26c39-0
00:02:11.840 --> 00:02:19.037
If we want to do research about, you know,
scale communication behaviors at scale,

c70532fc-751c-4a11-b841-26a5cff26c39-1
00:02:19.037 --> 00:02:24.760
we need a data set that reflects the
thing that we want to study.

3524c190-3e99-4880-8bed-60a7cf0b87ac-0
00:02:24.760 --> 00:02:27.840
And there aren't any or there hadn't been
any.

7c1e5cf7-ec15-4417-9ebc-02aae1aa575a-0
00:02:27.840 --> 00:02:29.760
There hadn't been any in the research
world.

546c4e52-9706-429f-a9d0-9a09998512ab-0
00:02:31.120 --> 00:02:35.613
But if I listen to the radio and so as
I'm trying to think about, well, OK,

546c4e52-9706-429f-a9d0-9a09998512ab-1
00:02:35.613 --> 00:02:39.930
how could we do some really innovative
research at scale, I thought, oh,

546c4e52-9706-429f-a9d0-9a09998512ab-2
00:02:39.930 --> 00:02:41.999
how about that story Corps archive?

3491c6a3-5257-425c-bc3a-4f9b8a27fde1-0
00:02:44.360 --> 00:02:48.280
And I,
I was just so grateful that it exists.

a586d43c-d550-4466-bcea-fa242f4abc27-0
00:02:48.560 --> 00:02:53.087
What we can do now is we have a
computational power where I can say, OK,

a586d43c-d550-4466-bcea-fa242f4abc27-1
00:02:53.087 --> 00:02:57.677
I want to feed in, you know, 100,
000 conversations and I want to feed in

a586d43c-d550-4466-bcea-fa242f4abc27-2
00:02:57.677 --> 00:02:59.599
all the words that people used.

c109c3a6-c917-4f6c-a4f0-82f900a0b771-0
00:02:59.600 --> 00:03:03.720
And maybe I can feed in who took which
turn when.

b95aca9b-8ecd-4bf1-be27-67e9d5331283-0
00:03:03.720 --> 00:03:06.227
And maybe I can also feed in,
you know what,

b95aca9b-8ecd-4bf1-be27-67e9d5331283-1
00:03:06.227 --> 00:03:08.400
what are their voice sound like, right?

870d0520-44c3-4a05-81eb-f45869883e92-0
00:03:08.400 --> 00:03:12.240
Were they stressed or, you know,
were there long pauses?

683dc861-ebfb-43de-aa81-dd7fd34788be-0
00:03:12.400 --> 00:03:16.863
You can feed in all that information,
which is the same information that humans

683dc861-ebfb-43de-aa81-dd7fd34788be-1
00:03:16.863 --> 00:03:18.760
are getting to make our decisions.

b4526e68-876e-428e-87bc-448ec5bad5fa-0
00:03:19.160 --> 00:03:23.636
But the computer can kind of do a better
job in a quantitative way,

b4526e68-876e-428e-87bc-448ec5bad5fa-1
00:03:23.636 --> 00:03:27.059
saying here are the biggest factors that,
you know,

b4526e68-876e-428e-87bc-448ec5bad5fa-2
00:03:27.059 --> 00:03:30.680
characterize one group in comparison to
another group.

97ff1095-c587-4379-a5de-8e68aee5a089-0
00:03:31.640 --> 00:03:36.991
Now that data is so much more plentiful
and that computational strategies have

97ff1095-c587-4379-a5de-8e68aee5a089-1
00:03:36.991 --> 00:03:41.259
allowed us to do more,
we can start to design research that is

97ff1095-c587-4379-a5de-8e68aee5a089-2
00:03:41.259 --> 00:03:46.882
more efficient and I think also allows us
to see things that we wouldn't otherwise

97ff1095-c587-4379-a5de-8e68aee5a089-3
00:03:46.882 --> 00:03:47.560
have seen.