O-15 Occupational Health: A Multi-Cohort Job Title Cleaning Project by Algorithm
Bibliographic record
Abstract
Introduction Occupational data in prospective cohort studies is often underutilized due to the human and financial resources required to code open-ended text, such as job titles. Recognizing the value of occupational data in health research, as well as potential errors associated with manual coding, an Automated Coding Algorithm (ACA)-NOC algorithm was developed utilizing a Natural Language Processing approach. Objectives We tested the ACA-NOC algorithm on two regional cohorts of a pan-Canadian cohort study, which represents the largest dataset an algorithm of this kind has been applied to. This process will harmonize and greatly expand the utility of the occupational data, enrich the research platforms, and further refine the efficiency of the algorithm. Methods The ACA-NOC algorithm was tested on data from the Canadian Partnership for Tomorrow’s Health (CanPath), a longitudinal cohort examining the role of genetic, environmental, lifestyle, and behavioural factors in the development of cancer and chronic disease. Using an iterative and interactive approach, the algorithm was applied to job title data from 111,000 questionnaires from two regional cohorts, coding the data to the Canadian National Occupation Classification (NOC) system. The algorithm was further refined based on each round of analysis, increasing the quantity of accurately coded data. Results Results from this research demonstrate the ability to refine the ACA-NOC algorithm with a 10% overall improvement in exact matching from the baseline algorithm. There were also instances where the algorithm performance was superior to the manual coding. The utilization of the algorithm offers significant savings in time, human resources and cost compared to a singular manual coding approach. Conclusions The coding and harmonization of this multi-cohort data demonstrates the value of the ACA-NOC algorithm, while increasing the utility of the CanPath data and research related to occupational health. Future research may involve comparisons between CanPath and international cohorts.
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 | 0.025 | 0.059 |
| Meta-epidemiology (narrow) | 0.001 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.003 |
| Bibliometrics | 0.003 | 0.003 |
| Science and technology studies | 0.002 | 0.001 |
| Scholarly communication | 0.003 | 0.001 |
| Open science | 0.003 | 0.003 |
| Research integrity | 0.001 | 0.002 |
| Insufficient payload (model declined to judge) | 0.010 | 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".