The day in conversation for 2026-11-03
discussion
Transcript
The discussion: Tom: That mediation analysis result regarding the monoclonal antibody therapy is interesting because if you only count the reduction in hospitalization risk when viral RNA is left-censored by assay limits, are we really missing a significant pathway, or is that bias inherent to the measurement process itself? Jane: I think it points toward a specific limitation in applying standard structural equation modeling here; the fractional imputation approach sounds promising for handling that censoring, but what assumptions about the relationship between viral load and clinical outcome are they making about the underlying distribution? Lu: From my perspective on mediation, if you’re using an EM algorithm to reduce bias, does that imply a specific functional form for the effect of the mediator itself, or is it agnostic to non-linear relationships in the pathway? Meng: I'm curious how that ties into what we saw last week with those sparse representation learning papers; is there any overlap in how you’re treating missing data versus how they are handling latent variable estimation when the observations themselves are incomplete due to assay limits? Lalam: If we look at the Beyond Vintage Rotation work, the oracle inference property sounds powerful, but that property relies on assuming
More episodes
- Daily Summary for 2026-11-19
- 2310.07973-Statistical Performance Guarantee for Subgroup Identification with Generic Machine Learning
- 2408.09155-Learning Robust Treatment Rules for Censored Data
- Daily Summary for 2026-11-19
- 2603.23397-Kinetic Langevin Splitting Schemes for Constrained Sampling
- 2606.09906-An information-geometric framework for mapping maximum potential biodiversity
- 2603.27718-Induced replication and the assessment of models
- 1307.0898-Entropy of a Zipfian Distributed Lexicon
- 2406.19974-Generalizing self-normalized importance sampling with couplings
- 2412.10213-Collaborative Optimal Design of Controlled Experiments with Subject Covariates