O1D.5 Non-detects in OSHA’s IMIS databank: are they short-term or shift-long samples?
Bibliographic record
Abstract
Objectives The Integrated Management Information System (IMIS) is the largest multi-industry source of exposure measurements available in North America. However, the lack of information on the censoring value (that depends on duration of sampling) of non-detected (ND) measurements considerably limits the usefulness of this databank. Released in 2010, the Chemical Exposure Health Database (CEHD) contains analytical results and measurement details, including duration of sampling for some of the records in IMIS. We assessed which ND results stored in IMIS are short-term (ST), and which are shift-long (LT) samples, based on information available in CEHD. Methods We analyzed exposure measurements for 54 agents from 1984–2009 (n=238,826). First, we calculated kappa coefficients (&_x0138;) for each agent to investigate the agreement between the exposure type of IMIS detected records (already indicated as ST or LT, i.e. selected by OSHA officers) and the exposure type suggested by sampling duration found in CEHD. If &_x0138; exceeded 0.3 for an agent, we employed classification and regression trees (CART) models to predict whether the ND results from IMIS should be classified as ST or LT samples. CART was developed using CEHD and applied to IMIS, relying on predictors common to both databanks: industry, reason for inspection, scope of inspection, region, union status, and year of sampling. Results The median proportion of ND results per agent was 37% (interquartile range (IQR)=22%–62%). The median &_x0138; was 0.45 (IQR=0.37–0.64) and 0.03 (IQR=0.01–0.16) for solvents/gases and metals/isocyanates, respectively. Solvents (n=22) and gases (n=7) were selected for CART modeling. Industry was the most important predictor variable in classifying ND results into either ST or LT. Conclusions This novel approach can be used to assign a censoring value to ND results, thus allowing more accurate inference about distribution of exposure levels in IMIS.
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 | 0.016 | 0.067 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.001 | 0.001 |
| Bibliometrics | 0.003 | 0.007 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.002 | 0.002 |
| Open science | 0.001 | 0.002 |
| Research integrity | 0.001 | 0.001 |
| Insufficient payload (model declined to judge) | 0.020 | 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".