Optimal designs for matching adjusted indirect comparison (MAIC)
Bibliographic record
Abstract
The thesis aimed to develop a method for optimizing design subject to matching the pre-defined baseline characteristics in clinical trials. As part of a clinical trial, a new treatment must be compared with a competitor treatment in order to determine its effect on the patient before the new treatment is launched. Ideally, we can directly compare the new treatment with competitor treatment in randomized controlled trials (RCTs). However, direct comparison is difficult to achieve due to various factors, such as time, price, regulation, and patents. A matching-adjusted indirect comparison (MAIC) method leverages all available data by adjusting average patient characteristics in trials with Individual patient data (IPD) to match those reported in the aggregate trials data (AgD). MAIC is a reweighting method in which the weights are calculated by deriving the propensity scores in the Individual patient’s data. This can reduce the bias. As IPD matches to the pre-defined baseline characteristics, we make use of optimal design theory and convert this into a constrained optimization problem. The Lagrangian method is used to determine the optimal design subject to satisfying the constraints of baseline characteristics. We formulate the Lagrangian and then transform the constrained problem to one where we simultaneously maximize several functions of the design weights. These functions have a common maximum of zero. In order to find the optimal design, we used the software R and a class of multiplicative algorithms. We then perform a sensitivity analysis and compare the Lagrangian method and the MAIC method by calculating the effective sample sizes (ESS). The higher the value of ESS the less information is lost due to reweighting. The Lagrangian method performs better than the MAIC method. Owing to the broad applicability of optimal design, we have tried to make use of this theory in order to obtain a better methodology in MAIC. The proposed methodology is quite flexible and can be applied to different types of constraints. The methodology can be applied to situation where there is a lack of direct comparison. It will also reduce the time and cost of running experiments.
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 | 0.121 | 0.269 |
| Meta-epidemiology (narrow) | 0.004 | 0.002 |
| Meta-epidemiology (broad) | 0.006 | 0.006 |
| Bibliometrics | 0.005 | 0.003 |
| Science and technology studies | 0.001 | 0.004 |
| Scholarly communication | 0.003 | 0.004 |
| Open science | 0.004 | 0.005 |
| Research integrity | 0.004 | 0.006 |
| Insufficient payload (model declined to judge) | 0.017 | 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".