BEGIN:VCALENDAR VERSION:2.0 PRODID:-//wp-events-plugin.com//7.4.0.1//EN TZID:Europe/Paris X-WR-TIMEZONE:Europe/Paris BEGIN:VEVENT UID:391@isdm.umontpellier.fr DTSTART;TZID=Europe/Paris:20250224T140000 DTEND;TZID=Europe/Paris:20250224T140000 DTSTAMP:20260828T114412Z URL:https://isdm.umontpellier.fr/events/doubly-robust-and-efficient-calibr ation-of-prediction-sets-for-censored-time-to-event-outcomes-3/ SUMMARY:Doubly Robust and Efficient Calibration of Prediction Sets for Cens ored Time-to-Event Outcomes DESCRIPTION:Online\n\nMachine Learning in Montpellier\, Theory &\; Pract ice\n\nOur objective is to construct well-calibrated prediction sets for a time-to-event outcome subject to right-censoring with guaranteed coverage . Our approach is inspired by modern conformal inference literature\, in t hat\, unlike classical frameworks\, we obviate the need for a well-specifi ed parametric or semi-parametric survival model to accomplish our goal. In contrast to existing conformal prediction methods for survival data\, whi ch restrict censoring to be of Type I\, whereby potential censoring times are assumed to be fully observed on all units in both training and validat ion samples\, we consider the more common right-censoring setting in which either only the censoring time or only the event time of primary interest is directly observed\, whichever comes first. Under a standard conditiona l independence assumption between the potential survival and censoring tim es given covariates\, we propose and analyze two methods to construct vali d and efficient lower predictive bounds for the survival time of a future observation. The proposed methods build upon modern semiparametric efficie ncy theory for censored data\, in that the first approach incorporates inv erse-probability-of-censoring weighting (IPCW)\, while the second approach is based on augmented-inverse-probability-of-censoring weighting (AIPCW). For both methods\, we formally establish asymptotic coverage guarantees\, and demonstrate both via theory and empirical experiments that AIPCW subs tantially improves efficiency over IPCW in the sense that its coverage err or bound is of second-order mixed bias type\, that is doubly robust\, and therefore guaranteed to be asymptotically negligible relative to the cover age error of IPCW.\n\nMachine Learning in Montpellier\, Theory &\; Prac tice ATTACH;FMTTYPE=image/jpeg:https://isdm.umontpellier.fr/wp-content/uploads/ 2026/06/ml-mtp-gC78d5.png CATEGORIES:ML MTP END:VEVENT BEGIN:VTIMEZONE TZID:Europe/Paris X-LIC-LOCATION:Europe/Paris BEGIN:STANDARD DTSTART:20241027T020000 TZOFFSETFROM:+0200 TZOFFSETTO:+0100 TZNAME:CET END:STANDARD END:VTIMEZONE END:VCALENDAR