BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//132.216.98.100//NONSGML kigkonsult.se iCalcreator 2.20.4//
BEGIN:VEVENT
UID:20251121T013140EST-4937Z2U0WI@132.216.98.100
DTSTAMP:20251121T063140Z
DESCRIPTION:Title: Causal Machine Learning Methods for Heterogeneous Treatm
 ent Effect Detection.\n\nAbstract: The conditional average treatment effec
 t (CATE) is frequently estimated to refute the homogeneous treatment effec
 t assumption. Under this assumption\, all units making up the population u
 nder study experience identical benefit from a given treatment. Uncovering
  heterogeneous treatment effects through inference about the CATE\, howeve
 r\, requires that covariates truly modifying the treatment effect be relia
 bly collected at baseline. CATE-based techniques will necessarily fail to 
 detect violations when effect modifiers are omitted from the data due to\,
  for example\, resource constraints. Severe measurement error has a simila
 r impact. To address these limitations\, we prove that the homogeneous tre
 atment effect assumption can be gauged through inference about contrasts o
 f the potential outcomes’ variances. We derive causal machine learning est
 imators of these contrasts and study their asymptotic properties. We estab
 lish that these estimators are doubly robust and asymptotically linear und
 er mild conditions\, permitting formal hypothesis testing about the homoge
 neous treatment effect assumptions even when effect modifiers are missing 
 or mismeasured. Numerical experiments demonstrate that these estimators’ a
 symptotic guarantees are approximately achieved in experimental and observ
 ational data alike. These inference procedures are then used to detect het
 erogeneous treatment effects in the re-analysis of a randomized controlled
  trial investigating targeted temperature management in cardiac arrest pat
 ients.\n\nVenue: UQAM Pavillon Président-Kennedy\, salle PK-5115\, Montréa
 l\n
DTSTART:20251127T203000Z
DTEND:20251127T213000Z
SUMMARY:Philippe Boileau (91ºÚÁÏÍø)
URL:/science/channels/event/philippe-boileau-mcgill-un
 iversity-369120
END:VEVENT
END:VCALENDAR