WEBVTT

00:00:09.490 --> 00:00:11.070
In this lecture,
we discuss the idea

00:00:11.070 --> 00:00:12.930
of predictive
analytics in medicine.

00:00:12.930 --> 00:00:14.690
Specifically, we
introduce the idea

00:00:14.690 --> 00:00:17.770
of using clustering methods
for better predicting heart

00:00:17.770 --> 00:00:20.090
attacks.

00:00:20.090 --> 00:00:23.510
Heart attacks are a common
complication of coronary heart

00:00:23.510 --> 00:00:27.120
disease, resulting from the
interruption of blood supply

00:00:27.120 --> 00:00:29.160
to part of the heart.

00:00:29.160 --> 00:00:30.620
Heat attack is the
number one cause

00:00:30.620 --> 00:00:33.830
of death for both men and
women in the United States.

00:00:33.830 --> 00:00:38.080
About one in every four
deaths is due to heart attack.

00:00:38.080 --> 00:00:41.110
A 2012 report from the
American Heart Association

00:00:41.110 --> 00:00:45.340
estimates about 715,000
Americans have a heart attack

00:00:45.340 --> 00:00:46.720
every year.

00:00:46.720 --> 00:00:48.680
To put this number
into perspective,

00:00:48.680 --> 00:00:51.220
this means that every
20 seconds, a person

00:00:51.220 --> 00:00:53.160
has a heart attack
in the United States.

00:00:53.160 --> 00:00:55.880
It is also equivalent
of September the 11th

00:00:55.880 --> 00:01:01.190
repeating itself every 24
hours, 365 days a year.

00:01:01.190 --> 00:01:02.840
Nearly half of
these attacks occur

00:01:02.840 --> 00:01:05.300
without prior warning signs.

00:01:05.300 --> 00:01:09.620
In fact, 250,000 Americans
die of sudden cardiac death

00:01:09.620 --> 00:01:14.710
yearly, which means 680
people every day die

00:01:14.710 --> 00:01:16.200
of sudden cardiac death.

00:01:19.360 --> 00:01:22.550
A heart attack has well-known
symptoms-- chest pain,

00:01:22.550 --> 00:01:26.770
shortness of breath,
upper body pain, nausea.

00:01:26.770 --> 00:01:28.340
The nature of heart
attacks, however,

00:01:28.340 --> 00:01:31.450
makes it hard to predict,
prevent, and even diagnose.

00:01:31.450 --> 00:01:33.380
Here are some statistics.

00:01:33.380 --> 00:01:36.000
25% of heart attacks are silent.

00:01:36.000 --> 00:01:39.900
47% of sudden cardiac deaths
occur outside hospitals,

00:01:39.900 --> 00:01:45.270
suggesting that many patients do
not act on early warning signs.

00:01:45.270 --> 00:01:49.250
Only 27% percent of
respondents to a 2005 survey

00:01:49.250 --> 00:01:57.190
recognized the symptoms
and called 911 for help.

00:01:57.190 --> 00:01:59.190
How can analytics help?

00:01:59.190 --> 00:02:01.390
The key to helping
patients is to understand

00:02:01.390 --> 00:02:03.430
the clinical
characteristics of patients

00:02:03.430 --> 00:02:06.700
in whom heart
attacks was missed.

00:02:06.700 --> 00:02:09.500
We need to better understand
the patterns in a patient's

00:02:09.500 --> 00:02:12.860
diagnostic history that
link to heart attack

00:02:12.860 --> 00:02:14.450
and to predicting
whether a patient is

00:02:14.450 --> 00:02:16.300
at risk for a heart attack.

00:02:16.300 --> 00:02:18.770
We'll see, in this
lecture, how analytics

00:02:18.770 --> 00:02:21.390
helps to understand
patterns of heart attacks

00:02:21.390 --> 00:02:25.160
and to provide good predictions
that in turn lead to improved

00:02:25.160 --> 00:02:29.280
monitoring and taking action
early and effectively.