Model specification, double robustness & targeted learning

20 open problems

Flexible learners and doubly robust estimators, and a precise account of what they do and do not protect against.

ID Problem Priority Verdict Maturity prank
MOD-01 Conventional STC reports a conditional effect at target means Very high Partially addressed Established 1
MOD-02 Convenience functional forms and untested flexible alternatives Very high Confirmed open Promising 1
MOD-03 Ecological versus within-study effect modification Very high Partially addressed Promising 1
MOD-05 No doubly robust PAIC estimator is indexed by bias parameters Very high Partially addressed Established 1
MOD-06 No TMLE or RMST-targeted estimator for population-adjusted transport Very high Overstated Promising 1
MOD-07 No influence-function theory for mixed-granularity multi-arm networks Very high Confirmed open Promising 1
MOD-09 No criterion says when ML-NMR should be preferred to MAIC or STC Very high Partially addressed Established 1
MOD-13 No reporting convention exposes prior-dominated contrasts in PAIC Very high Partially addressed Emerging 1
MOD-19 No valid selection criterion for cut points and powers in time-varying survival NMA Very high Confirmed open Promising 1
MOD-04 No post-selection inference for model choice in population adjustment High Confirmed open Promising 2
MOD-08 Cross-fitting folds and rate conditions undefined for PAIC High Partially addressed Promising 2
MOD-12 Mixture-weight calibration is unvalidated for subnetwork bridges High Partially addressed Promising 2
MOD-14 Network meta-interpolation has no stated estimand and no implementation High Partially addressed Emerging 2
MOD-15 No calibrated rule says when within- and across-trial interactions may be pooled High Confirmed open Promising 2
MOD-16 No criterion chooses pooled or arm-separate matching in anchored MAIC High Confirmed open Established 2
MOD-17 No selection-adjusted interval for the treatment a network ranks best High Confirmed open Emerging 2
MOD-20 No criterion or error-based evidence selects arm-synthesis over contrast-synthesis High Confirmed open Established 2
MOD-10 Flexible interaction surfaces inside network evidence synthesis Medium-high Confirmed open Promising 3
MOD-11 No mixture-over-populations model transports to a declared target Medium-high Partially addressed Emerging 3
MOD-18 Risk-modeling network meta-regression treats a fitted risk score as observed Medium-high Confirmed open Emerging 3
No matching items

Back to the full catalog