BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//talks.cam.ac.uk//v3//EN
BEGIN:VTIMEZONE
TZID:Europe/London
BEGIN:DAYLIGHT
TZOFFSETFROM:+0000
TZOFFSETTO:+0100
TZNAME:BST
DTSTART:19700329T010000
RRULE:FREQ=YEARLY;BYMONTH=3;BYDAY=-1SU
END:DAYLIGHT
BEGIN:STANDARD
TZOFFSETFROM:+0100
TZOFFSETTO:+0000
TZNAME:GMT
DTSTART:19701025T020000
RRULE:FREQ=YEARLY;BYMONTH=10;BYDAY=-1SU
END:STANDARD
END:VTIMEZONE
BEGIN:VEVENT
CATEGORIES:Isaac Newton Institute Seminar Series
SUMMARY:Estimation of Causal Effects in Network-Dependent
Observational Data - Oleg Sofrygin (University of
California\, Berkeley)
DTSTART;TZID=Europe/London:20160712T153000
DTEND;TZID=Europe/London:20160712T160000
UID:TALK66719AThttp://talks.cam.ac.uk
URL:http://talks.cam.ac.uk/talk/index/66719
DESCRIPTION:Co-author: Mark J. van der Laan (University
of California\, Berkeley\, CA)
We outline the framework of targeted maximum l
ikelihood estimation (TMLE) in observational netw
ork data. Consider a dataset in which each observa
tional unit is causally connected to other units
via a known social or geographical network. For e
ach unit we observe their baseline covariates\, th
eir exposure and their outcome\, and we are inter
ested in estimating the effect of a single time-po
int stochastic intervention. We propose a semi-pa
rametric statistical model that allows for betwee
n-unit dependencies: First\, unit-level exposure c
an depend on the baseline covariates of other con
nected units. Second\, the unit-level outcome can
depend on the baseline covariates and exposures o
f other connected units. We impose some restricti
ons on our model\, e.g.\, assuming that the unit
39\;s exposure and outcome depend on other units
as some known (but otherwise arbitrary) summary m
easures of fixed dimensionality. A practical appli
cation of our approach is demonstrated in a large
-scale networ k simulation study that applies two
newly developed R packages: simcausal and tmlenet
. We also discuss some extensions of our work tow
ards estimation in longitudinal data.
LOCATION:Seminar Room 1\, Newton Institute
CONTACT:INI IT
END:VEVENT
END:VCALENDAR