Source identification for multiple chemical exposure using pattern recognition and classification techniques.
Environ Sci Technol 1993 Nov; 27(12):2430-2434
The use of pattern recognition techniques for characterizing occupational exposure to organic solvents in a printing/reproduction facility was described. The facility included shops for printing, photolithography, binding, and polycopy. The study group consisted of 18 exposed workers; five workers from other parts of the same building served as controls. Breathing zone samples were collected during normal working hours on each day for 1 week using passive samplers and charcoal tubes. Principal component analysis (PCA) of the printing air samples produced two statistically significant components. Soft independent modeling of class analogy (SIMCA) accurately classified 95% of the samples. Using the same air exposure data, classification and regression trees (CART) analysis correctly classified 91% of the cases. PCA plus CART analysis identified sources of exposure and verified that exposure classification of workers by job type corresponded closely to the exposures measured using personal sampling. The authors note that the usefulness of this system over other techniques is both an improved identification of the sources of exposure and an improved characterization of how these sources interact at the point of the individual. The authors conclude that pattern recognition techniques are easily able to identify characteristic patterns of mixed chemical exposure. Since these patterns of exposure are based on actual multiple measurements of exposure, they therefore provide a more accurate classification of exposure than job classification does. Finally, classification becomes less effective as exposures approach the limits of sampling and analysis. The technique can be used to successfully reduce multivariate exposure measurement to a few summary variables.
NIOSH-Publication; NIOSH-Grant; Grants-other; Air-quality-monitoring; Organic-solvents; Indoor-air-pollution; Air-sampling; Chemical-analysis; Analytical-chemistry; Analytical-methods; Printing-industry; Publishing-and-printing; Indoor-environmental-quality
Biomedical & Environ Hlth Scis University of California 322 Warren Hall Berkeley, Calif 94720
Environmental Science and Technology
University of California Berkeley, Berkeley, California