The Industrial Hygiene Data Standard, Artificial Intelligence, and Exposure Modelling
By Christopher A. Smith and Dr. Trevor Runcie
The Industrial Hygiene Data Standard is designed to promote the quality, consistency, and interoperability of industrial hygiene (IH) data. It provides a common framework for organizations to capture, store, and exchange industrial hygiene monitoring data in a structured and consistent manner, regardless of the technology or tools used. It ensures that industrial hygiene data can be easily shared, analyzed, and compared across applications and organizations, enabling meaningful insights and informed decision-making.
There are many models for exposure prediction. Typically, these models “calculate” an estimated exposure using parameters such as workers’ proximity to the exposure source, the scale of the operation, the specific task being performed, and the engineering control measures that have been implemented. Over time, these models can be refined using real monitoring results.
An alternative method for predicting exposures is the use of data to derive relationships between exposure attributes and real monitoring results. The constant stream of articles and announcements about artificial intelligence (AI) and Big Data has created substantial interest in this topic in the IH community, but the community must address several problems related to IH data and monitoring before AI and Big Data solutions are widely adopted.
Primarily, very few organizations, if any, have recorded sufficient data with adequate quality and consistency to deliver the insights that AI and Big Data seem to promise. By adopting a common data standard, even small organizations can choose to pool their data sets with similar organizations to grow the shared knowledge base. Comparability and benchmarking are critical—organizations need to compare data across different industries, regions, and time periods to identify patterns and trends and establish benchmarks for best practices.
Additionally, for most data sets, monitoring does not capture adequate “context information,” which is fundamental to AI and modeling. If only compliance data is recorded—such as who was sampled, for what substance, and when—these data will never be sufficient to predict novel exposures. If data sets also record parameters such as the type of task, the workers’ proximity to the source of exposure, the size of the room, and the room’s ventilation rate, it becomes possible to explore relationships between these elements and the monitoring results.
In 1996, AIHA and the American Conference of Governmental Industrial Hygienists (ACGIH) collaborated on a joint task group that made recommendations for data to be recorded in IH databases. The task group developed what was, for the time, a comprehensive list of data elements that included 134 total variables. Unfortunately, the task group’s guidance was not widely adopted in the industry. If it had, there would be more data suitable for use in AI systems now.
The intention of the Industrial Hygiene Data Standard is that adopters will be able to leverage other data sets using the same framework, regardless of whether they use a professional IH database, a system developed in-house, a spreadsheet, or even paper. This will mean that the promise of AI and Big Data can be delivered to the entire IH community.
Christopher A. Smith is an innovative, analytic thinker and environmental chemist / toxicologist. He has more than 28 years of experience providing scientific data analysis and novel development approaches to addressing chemical hazard, risk, and information management. His focus is decision support, systems design, and software development. Chris holds a Master of Environmental Management (MEM) from Duke where he studied computational toxicology. He is an avid thru-hiker, mountaineer, and farmer (Clouds Rest Acres).
Dr. Trevor Runcie is a scientist and entrepreneur with over 30 years of experience designing and building industrial hygiene software solutions. He has recently completed a Master's degree in Health Data at Cambridge University in England and chose to focus his dissertation on issues around IH data standards, IH data analytics, and IH data modeling. Trevor also has a Master's degree in AI, and his PhD is in Computer Science.
About Redshift Technologies
Redshift Technologies is a custom software development firm specializing in data-driven websites and business applications. Our customers include Fortune 500 companies, medical organizations, and technology start-ups in sectors including consumer products, pharmaceuticals, and financial services. Based in New York City since 1996, we design and program virtually anything our customers describe and can leverage our existing tools for sustainability, compliance, and risk analysis.