-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathresearch.html
executable file
·96 lines (79 loc) · 3.31 KB
/
research.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; charset=iso-8859-1">
<TITLE>Research Interests</TITLE>
</HEAD>
<BODY BGCOLOR="white" TEXT="#000000" >
<TABLE BORDER=0 CELLSPACING=8 CELLPADDING=0 WIDTH="100%" >
<TR>
<TD bgcolor="#DDDDDD" VALIGN=TOP WIDTH="18%">
<P><A HREF="index.html">Home</A>
<P>Research
<P><A HREF="teaching.html">Teaching</A>
<P><A HREF="pubs.html">Publications</A>
<P><A HREF="personal.html">Personal info</A>
<P><A HREF="altindex.html">Favorite links</A>
</TD>
<TD VALIGN=TOP colspan=2>
<H3>Research Interests </H3>
<p>
I am interested in statistical issues related to public health.
Much of my current research focuses on the design of epidemiologic and
clinical studies, and associated questions of data access
such as security and privacy, extraction from large
central registries, and
distributed statistical computing.
I was introduced to the
area through collaborative outcomes research with the
Back Pain Outcomes Assessment Team at the University of Washington.
An important focus of my research is cost-effective study design.
Recently, two-phase or response-selective approaches
have been proposed as economical. The case-control
study in which cases are oversampled relative to their overall
population frequency may be viewed as an example, and is known to be
far more efficient than a cohort study when the disease is rare.
More generally, the response and a set of inexpensive covariates can
be measured on all subjects. Then, in a second phase, those with rare
responses and/or exposures can be oversampled for a more informative and
costly covariate, while a reduced fraction of individuals with common
responses and exposures can be sampled. Thus, a
subset of subjects will have data <I> missing by design </I> on the
costly covariate. I have done some work on large sample theory for
maximum likelihood in two-phase studies with
<a href=http://www.stat.washington.edu/jaw/>Jon Wellner</a> and
<a href=http://www.biostat.washington.edu/~norm/>Norm Breslow</a>.
The work carries over naturally to
more general missing-data problems in the statistical analysis of
clinical trial data. With <a href="http://www.stat.ncsu.edu/~tsiatis/">
Dr. A. (Butch) Tsiatis </a> (<a href="http://www.stat.ncsu.edu">
North Carolina
State University </a>), I am investigating efficient estimation and testing
in semiparametric models with data that are missing by chance.
<p>
I also have research interests in HIV vaccine development.
Given the scale of the public health problem HIV poses in the
developing world, the need for effective vaccines is clear.
In a project with
<a href="http://www.bios.unc.edu/people/faculty/show.phtml?id=1045">
Dr. Francoise Seillier-Moiseiwitsch </a>
(<a href=""http://www.bios.unc.edu/> University of North Carolina at
Chapel Hill </a>) and
<a href="http://www.math.sfu.ca/~jgraham"> Dr. Jinko Graham </a>
A
(<a href=""http://www.math.sfu.ca/"> Simon Fraser University</a>),
we are using HIV
envelope sequences collected from a sample of Kenyan women
to help assess the number and diversity of HIV quasispecies
transmitted to an individual. The larger the number of transmitted
quasispecies, the more difficult the task of developing an effective
vaccine.
<p>
<hr size=3 width="100%">
<I>
Last modified: Fri Jun 30 10:51:54 PDT 2000
</I>
</TD>
</TR>
</TABLE>
</BODY>
</HTML>