BEGIN:VCALENDAR
VERSION:2.0
PRODID:-//Talks.cam//talks.cam.ac.uk//
X-WR-CALNAME:Talks.cam
BEGIN:VEVENT
SUMMARY:Robust machine learning for causal inference in health care - Davi
 d Sontag\, MIT
DTSTART:20190327T110000Z
DTEND:20190327T120000Z
UID:TALK120847@talks.cam.ac.uk
CONTACT:Adrian Weller
DESCRIPTION:Electronic health records are now pervasive\, presenting an in
 credible opportunity to use retrospective data to learn about medicine and
  to improve health care. Machine learning can help answer questions such a
 s\, "What conditions does this patient have?"\, "When will this patient's 
 disease progress?" and "How should we optimally treat this disease?". Prop
 erly answering these questions requires tackling head-on questions of caus
 ality\, specifically how to infer causality from high-dimensional observat
 ional data. Machine learning and causal inference in health care introduce
 s additional challenges including little labeled data\, significant missin
 g data\, censoring\, and the need to characterize individual-level uncerta
 inty. I will discuss several new methodologies that my group has created t
 o address these challenges\, with a particular focus on disease progressio
 n modeling and estimation of individual treatment effect. Specifically\, I
  discuss provable guarantees for causal inference under model misspecifica
 tion (Johansson et al. ICML '16\, Shalit et al. ICML '17)\, approaches for
  causal inference with unobserved confounding (Louizos et al. NeurIPS '17)
 \, how to check assumptions for off-policy reinforcement learning (Gottesm
 an et al. Nature Medicine '19\, Oberst et al. '19)\, assessing overlap (Jo
 hansson et al.\, '19)\, and learning nonlinear dynamical models using the 
 deep Markov model (Krishnan et al.\, AAAI '17).
LOCATION:Engineering Department\, CBL Room BE-438.
END:VEVENT
END:VCALENDAR