Design considerations in an active medical product safety monitoring system
Bibliographic record
Abstract
Active medical product monitoring systems, such as the Sentinel System, will utilize electronic healthcare data captured during routine health care. Safety signals that arise from these data may be spurious because of chance or bias, particularly confounding bias, given the observational nature of the data. Applying appropriate monitoring designs can filter out many false-positive and false-negative associations from the outset. Designs can be classified by whether they produce estimates based on between-person or within-person comparisons. In deciding which approach is more suitable for a given monitoring scenario, stakeholders must consider the characteristics of the monitored product, characteristics of the health outcome of interest (HOI), and characteristics of the potential link between these. Specifically, three factors drive design decisions: (i) strength of within-person and between-person confounding; (ii) whether circumstances exist that may predispose to misclassification of exposure or misclassification of the timing of the HOI; and (iii) whether the exposure of interest is predominantly transient or sustained. Additional design considerations include whether to focus on new users, the availability of appropriate active comparators, the presence of an exposure time trend, and the measure of association of interest. When the key assumptions of self-controlled designs are fulfilled (i.e., lack of within-person, time-varying confounding; abrupt HOI onset; and transient exposure), within-person comparisons are preferred because they inherently avoid confounding by fixed factors. The cohort approach generally is preferred in other situations and particularly when timing of exposure or outcome is uncertain because cohort approaches are less vulnerable to biases resulting from misclassification.
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 | 0.087 | 0.091 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.001 | 0.001 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.004 | 0.003 |
| Open science | 0.003 | 0.002 |
| Research integrity | 0.004 | 0.003 |
| Insufficient payload (model declined to judge) | 0.011 | 0.003 |
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".