Beyond Crosswalks: Reliability of Exposure Assessment Following Automated Coding of Free-Text Job Descriptions for Occupational Epidemiology
Bibliographic record
Abstract
Epidemiologists typically collect narrative descriptions of occupational histories because these are less prone than self-reported exposures to recall bias of exposure to a specific hazard. However, the task of coding these narratives can be daunting and prohibitively time-consuming in some settings. The aim of this manuscript is to evaluate the performance of a computer algorithm to translate the narrative description of occupational codes into standard classification of jobs (2010 Standard Occupational Classification) in an epidemiological context. The fundamental question we address is whether exposure assignment resulting from manual (presumed gold standard) coding of the narratives is materially different from that arising from the application of automated coding. We pursued our work through three motivating examples: assessment of physical demands in Women's Health Initiative observational study, evaluation of predictors of exposure to coal tar pitch volatiles in the US Occupational Safety and Health Administration's (OSHA) Integrated Management Information System, and assessment of exposure to agents known to cause occupational asthma in a pregnancy cohort. In these diverse settings, we demonstrate that automated coding of occupations results in assignment of exposures that are in reasonable agreement with results that can be obtained through manual coding. The correlation between physical demand scores based on manual and automated job classification schemes was reasonable (r = 0.5). The agreement between predictive probability of exceeding the OSHA's permissible exposure level for polycyclic aromatic hydrocarbons, using coal tar pitch volatiles as a surrogate, based on manual and automated coding of jobs was modest (Kendall rank correlation = 0.29). In the case of binary assignment of exposure to asthmagens, we observed that fair to excellent agreement in classifications can be reached, depending on presence of ambiguity in assigned job classification (κ = 0.5-0.8). Thus, the success of automated coding appears to depend on the setting and type of exposure that is being assessed. Our overall recommendation is that automated translation of short narrative descriptions of jobs for exposure assessment is feasible in some settings and essential for large cohorts, especially if combined with manual coding to both assess reliability of coding and to further refine the coding algorithm.
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 | 0.126 | 0.375 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.004 | 0.004 |
| Science and technology studies | 0.001 | 0.002 |
| Scholarly communication | 0.005 | 0.005 |
| Open science | 0.002 | 0.005 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 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".