[Statlist] Research Seminar in Statistics | *FRIDAY 24 FEBRUARY 2023* | GSEM, University of Geneva

[gsem-support-instituts]

Dear all,

We are pleased to invite you to our next Research Seminar, organized by Professor Sebastian Engelke on behalf of the Research Center for Statistics. (https://www.unige.ch/gsem/en/research/institutes/rcs/team/)

FRIDAY 24 FEBRUARY 2023 at 11:15 am, Uni Mail M 3393 & Online
Zoom research webinar: https://unige.zoom.us/j/92924332087?pwd=U1U1NFk4dTFCRHBMeWYrSDBQcXBiQT09
Meeting ID: 929 2433 2087
Passcode: 399192

Optimal Regimes for Algorithm-Assisted Human Decision-Making
Aaron Leor SARVET, EPFL, Switzerland
https://people.epfl.ch/aaron.sarvet?lang=en

ABSTRACT:
Foundational work on causal inference and dynamic treatment regimes presents a promising pathway towards precision medicine. In a precision-medicine system, decision rules might be algorithmically individualized based on an optimal rule previously learned from non-experimental or experimental data. However, there is some resistance to the notion that implementation of an optimal regime, successfully learned from the data, will result in better expected outcomes on average, compared to existing human-decision rules: care providers may be inclined to override the treatment recommendations provided by the identified optimal regimes, based on their privileged patient observations. In this talk, I will review existing methodology for learning optimal regimes and clarify the validity of the care provider's skepticism. Then, I will present methodology for leveraging human intuition by identifying a super-optimal regime using data generated by either nonexperimental or experimental studies, and clarify when a fusion of such data is beneficial. The superoptimal regime will indicate to a care provider -- in an algorithm-assisted decision setting -- precisely when expected outcomes would be maximized if the care provider would override the optimal regime recommendation and, importantly, when the optimal regime recommendation should be followed regardless of the care-provider's assessment. If time allows, I will also discuss recent proposals for incorporating the Hippocratic "first, do no harm" in optimal regime learning, and discuss their practical and philosophical implications.


> View the Research Seminar agenda: https://www.unige.ch/gsem/en/research/seminars/rcs/

Regards,


Marie-Madeleine

Marie-Madeleine Novo
Assistant to the Research Institutes
Mail.unige.ch