1
00:00:09,490 --> 00:00:13,220
Let us examine what we
learned in this lecture.

2
00:00:13,220 --> 00:00:17,790
We built an expert-trained
model by a physician

3
00:00:17,790 --> 00:00:21,920
that can accurately identify
diabetic patients receiving

4
00:00:21,920 --> 00:00:23,410
low quality care.

5
00:00:23,410 --> 00:00:26,930
We have observed that the out
of sample accuracy of the model

6
00:00:26,930 --> 00:00:28,730
was 78%.

7
00:00:28,730 --> 00:00:32,600
But most importantly, the
model identifies most patients

8
00:00:32,600 --> 00:00:36,520
receiving poor care, which
is the major objective

9
00:00:36,520 --> 00:00:37,220
in the study.

10
00:00:39,780 --> 00:00:42,980
Logistic regression
models provide

11
00:00:42,980 --> 00:00:48,970
probabilities of somebody
receiving poor quality care.

12
00:00:48,970 --> 00:00:53,630
These probabilities can be
used to prioritize patients

13
00:00:53,630 --> 00:01:00,920
for intervention, a particularly
useful outcome from the study.

14
00:01:00,920 --> 00:01:05,860
While the accuracy is
reasonably high, 78%,

15
00:01:05,860 --> 00:01:08,120
it can be, of course,
further improved.

16
00:01:08,120 --> 00:01:13,530
In that respect, I expect that
electronic medical records, not

17
00:01:13,530 --> 00:01:16,400
only claims, could
be used in the future

18
00:01:16,400 --> 00:01:19,590
to enhance the predictive
capability of such models.

19
00:01:19,590 --> 00:01:25,100
So a model like the
one we built can

20
00:01:25,100 --> 00:01:30,090
be used to analyze literally
millions of records.

21
00:01:30,090 --> 00:01:35,620
Whereas a human can only
accurately analyze rather small

22
00:01:35,620 --> 00:01:37,440
amounts of information.

23
00:01:37,440 --> 00:01:43,039
So clearly such a model
allows significantly larger

24
00:01:43,039 --> 00:01:43,539
scalability.

25
00:01:46,800 --> 00:01:51,289
Of course models do not
replace expert judgement.

26
00:01:51,289 --> 00:01:59,880
However, models provide a way
to translate expert judgement

27
00:01:59,880 --> 00:02:04,910
to a reproducible, testable
prediction methodology that

28
00:02:04,910 --> 00:02:10,340
has significantly higher
scalability, as we discussed.

29
00:02:10,340 --> 00:02:14,490
And of course experts can
continuously improve and refine

30
00:02:14,490 --> 00:02:17,590
the model, as we have
seen in this lecture.

31
00:02:21,420 --> 00:02:24,680
Finally, and quite
importantly, models

32
00:02:24,680 --> 00:02:28,050
can integrate assessments
of multiple experts

33
00:02:28,050 --> 00:02:31,590
into one final, unbiased,
and unemotional prediction.

34
00:02:34,180 --> 00:02:38,020
And such methods of combining
assessments and combining

35
00:02:38,020 --> 00:02:41,880
models is a tool
that we will use

36
00:02:41,880 --> 00:02:44,550
later in the class
on multiple occasions

37
00:02:44,550 --> 00:02:51,079
as a way of enhancing and
improving quantitative models.