Automatic approaches to clustering occupational description data for prediction of probability of workplace exposure to beryllium

Abstract

We investigated automatic approaches for clustering data that describes occupations related to hazardous airborne exposure (beryllium). The regulatory compliance data from Occupational Safety and Health Administration includes records containing short free text job descriptions and associated numerical exposure levels. Researchers in public health domain need to map job descriptions to Standard Occupational Classification (SOC) nomenclature for estimating occupational health risks. Previous manual process was time-consuming and did not advance so far to linkage to SOC. We investigated alternative automatic approaches for clustering job descriptions. The clustering results are the first essential step towards discovery of corresponding SOC terms. Our study indicated that the Tolerance Rough Set with Jaccard similarity was a better combination overall. The utility of the algorithm was further verified by applying logistic regression and validating that the predictive power of the automatically generated classifications, in terms of association of "job" with probability of exposure to beryllium above certain threshold, closely approached that of the manually assembled classification of the same 12,148 records.

Workplace Measurements by the US Occupational Safety and Health Administration since 1979: Descriptive Analysis and Potential Uses for Exposure Assessment

“…Some authors have manually recoded this variable when their dataset was restricted to few industries (Teschke et al, 1999;Hamm and Burstyn, 2011). More recently, Slutsky et al developed an algorithm to automatically create standard occupations across all industries in IMIS from the text description (Slutsky et al, 2011). The analysis of variables internal to IMIS, while informative, cannot evaluate adequately the relationship between exposure levels in IMIS and those occurring in US workplaces.…”

Women’s occupational exposure to polycyclic aromatic hydrocarbons and risk of breast cancer

“…Industries were classified using the North American Industry Classification System 2007 (Canadian edition) and occupations were classified using the Standard Occupation Classification 2010 (US edition). Industrial classification was done manually, while occupational classification was completed using an automated approach,24 which has previously been assessed for its reliability,25 and then manually reviewed by the authors (DGL, IB) to ensure accuracy. The JEM estimates the probability of job-specific exposure (τ) exceeding the permissible exposure level (PEL=0.2 mg/m 3 ) for PAHs (θ=Pr(τ>PEL)).…”

Beyond Crosswalks: Reliability of Exposure Assessment Following Automated Coding of Free-Text Job Descriptions for Occupational Epidemiology

“…We previously developed an automated method to cluster free-text descriptions of occupations into groups with similar meaning and applied the method to identify jobs with different probability of exposure to beryllium, in the context of exposure measurements that were collected by Occupational Safety and Health Administration (OSHA) in the Integrated Management Information System (IMIS) (Slutsky et al, 2011). Compared to the crosswalk created in manual coding of free-text descriptions of occupations (Hamm and Burstyn, 2011), the automated coding produced clusters with a similar ability to discriminate among jobs with different exposures, leading us to conclude that the automated coding approach is promising and produces results that are no worse than those by human coders at the fraction of cost and effort (Slutsky et al, 2011). However, such findings have limited transferability to other settings because the clustered occupations did not correspond to widely used systems for coding occupations, such as the major occupation groups used by the SOC (Bureau of Labour and Statistics, 2013).…”