Toward the Estimation of Unbiased Disease Prevalence Estimates Using Administrative Health Records
Bibliographic record
Abstract
Data are information. And what we do with that information, how we process it, and interpret it can be complicated. It should not come as a surprise that these days there is a lot of talk about “big data” – about its promise, its potential, and its pitfalls. Big data (e.g., administrative, birth certificates, claims, electronic health records, registers) are growing in size, accessibility, and application. However, repurposing data from their original use to the research environment requires careful attention. Truthfully, whether we are talking about statistical analysis of small clinical datasets or supervised learning algorithms in big datasets, some of the same principles apply. No matter what, understanding where our data come from informs our design, our analysis, and most importantly, our interpretation. There are 3 major sources of bias that determine whether inferences from a dataset are a close approximation of the truth: confounding, selection, and information. Confounding occurs when an association between 2 factors can be explained by an (often unmeasured) extraneous factor. Confounding often limits our ability to make truthful inferences about causality. Selection bias may occur when the choice of dataset limits the ability to generalize findings to the population affected by a disease. For example, using only drug claims data or hospitalization data to infer the prevalence of osteoarthritis (OA) may underestimate the condition because there may be individuals who may not need medication or have not been hospitalized in the time window evaluated. Information bias (often referred to as misclassification or … Address correspondence to J.F. Simard, Assistant Professor, Division of Epidemiology, Department of Health Research and Policy, Stanford School of Medicine, HRP Redwood Building, Room T152, 259 Campus Drive, Stanford, California 94305-5405, USA. E-mail: jsimard{at}stanford.edu
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 | 0.101 | 0.291 |
| Meta-epidemiology (narrow) | 0.002 | 0.003 |
| Meta-epidemiology (broad) | 0.004 | 0.003 |
| Bibliometrics | 0.009 | 0.007 |
| Science and technology studies | 0.001 | 0.004 |
| Scholarly communication | 0.006 | 0.005 |
| Open science | 0.005 | 0.005 |
| Research integrity | 0.003 | 0.006 |
| Insufficient payload (model declined to judge) | 0.002 | 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".