Bibliographic record
Abstract
Semiparametric efficient methods in causal inference have been developed to robustly and efficiently estimate causal parameters. As in general causal estimation, the methods rely on a set of mathematical assumptions that translate into requirements of causal knowledge and confounder identification. Targeted maximum likelihood estimation (TMLE) methodology has been developed as a potential improvement on efficient estimating equations, in that it shares the qualities of double robustness (unbiasedness under partial misspecification) and semiparametric efficiency, but can be constructed to provide boundedness of parameter estimates, robustness to data sparsity, and a unique estimate.This thesis, composed primarily of three manuscripts, presents new research on the analysis of longitudinal and survival data with time-dependent confounders using TMLE. The first manuscript describes the construction of a two time-point TMLE using a generalized exponential distribution family member as the loss function for the outcome model. It demonstrates the robustness of the continuous version of this TMLE algorithm in a simulation study, and uses a modified version of the method in a simplified analysis of the PROmotion of Breastfeeding Intervention Trial (PROBIT) where evidence for a protective causal effect of breastfeeding on gastrointestinal infection is obtained.The second manuscript presents a description of several substitution estimators for longitudinal data, a specialized implementation of a longitudinal TMLE method, and a case study using the full PROBIT dataset. The K time point sequential TMLE algorithm employed (theory previously developed), implemented nonparametrically using Super Learner, differs fundamentally from the strategy used in the first manuscript, and offers some benefits in computation and ease of implementation. The analysis compares different durations of breastfeeding and the related exposure-specific (and censoring-free) mean counts of gastrointestinal infections over the first year of an infant's life and concludes that a protective effect is present. Simulated data mirroring the PROBIT dataset was generated, and the performance of TMLE was again assessed.The third manuscript develops a methodology to estimate marginal structural models for survival data. Utilizing the sequential longitudinal TMLE algorithm to estimate the exposure-specific survival curves for all exposure patterns, it demonstrates a way to combine inference in order to model the outcome using a linear specification. This article presents the theoretical construction of two different types of marginal structural models (modeling the log-odds survival and the hazard) and presents a simulation study demonstrating the unbiasedness of the technique. It then describes an analysis of the Canadian Co-infection Cohort study undertaken with one of the TMLE methods to fit survival curves and a model for the hazard function of development of end-stage liver disease (ESLD) conditional on time and clearance of the Hepatitis C virus.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.015 | 0.056 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.002 | 0.002 |
| Bibliometrics | 0.002 | 0.003 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.003 |
| Open science | 0.003 | 0.003 |
| Research integrity | 0.002 | 0.004 |
| Insufficient payload (model declined to judge) | 0.005 | 0.002 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".