Artificial Intelligence-Based Asbestos Testing: Bringing a New Solution to an Old Problem

Ryan Fisher

With every year of scientific pursuit and research, it seems that we discover an increasing number of hazardous materials and potentially lethal illnesses linked to them. Take, for instance, microplastics—a relatively new discovery in health science that has been linked to a variety of developing ailments. While this science is progressing, it often leaves longer-known issues by the wayside. One great example is asbestos. 

Asbestos is a term used to describe a group of minerals that were popularly used in building materials and commercial products during the 20th century. Of its most common uses, asbestos was popularly added to building materials such as insulation, mastics, roofing, and vinyl flooring. Decades after it became a popular facet of American commerce, it was found to be a hazardous material, particularly when it releases microscopic mineral fibers when disturbed. 

Nowadays, it is common knowledge that asbestos can cause a range of diseases, including mesothelioma, asbestosis, and lung cancer, to name a few. Since the discovery of its adverse health effects, the American government has put heavy regulations on the material, creating a booming industry of asbestos abatement and testing services.

Asbestos Testing is a Relatively Limited Practice

Interestingly enough, the science behind asbestos testing has remained the same since it was popularized in the 1970s. The process usually utilizes polarized light microscopy to identify fibers in bulk building materials and phase contrast microscopy to identify fibers in indoor air samples. For both methods, analytical procedures are slow, time-intensive, and rely on the accuracy and training of individual laboratory analysts. As you can guess, this can create a bottleneck of lab work and lead to incorrect readings in some cases. Likewise, the training and expertise required from analysts often means that asbestos testing can cost consumers and contractors a pretty penny, adding to its inaccessibility to some. 

While some laboratories have adopted digital systems for sample preparation and intake, there has been little effort to update or improve the process of analyzing asbestos.

Asbestos Meets the Digital Age – Phone Applications Improve Screening Processes

Luckily, there have been some innovators who have sought to combine the worlds of asbestos testing and technology. Early on, many groups proposed digital screening processes to help start the process of screening for and identifying possible asbestos-containing materials. Some of these programs had incredible success stories, allowing people from across the globe to discover, track, and flag materials to be safely removed from public and private spaces. 

One incredibly innovative program took the form of a smartphone application. The app utilized a detailed questionnaire, which included questions regarding the location of a material, the type of material, and any possible manufacturing information to help identify if a material was likely to contain asbestos fibers. Paired with a comprehensive digital database of countless known asbestos-containing materials, the application allowed people to get a reasonably accurate screening of materials that could contain asbestos at the palm of their hands. 

In essence, efforts such as this marked the incredibly nuanced pairing of new-age solutions to age-old problems.

Artificial Intelligence and Machine Learning: A Solution to Quick Asbestos Fiber Counting

Others sought to attack the issues of asbestos with technology more directly, designing computer-learning-based methods to create digital workflows of asbestos testing. As mentioned previously, asbestos testing typically relies on traditional microscope-based techniques that rest upon the skill of training of each analyst. In addition to the time required for proper training, these techniques mean that each sample must be individually analyzed for asbestos, usually resulting in bottlenecked laboratory workflows and increased costs. 

Work done by the research Group of Lee et al. sought to create an innovative solution to this problem, employing the use of artificial intelligence and machine learning. Specifically, they created algorithms to teach computer systems how to read air samples for asbestos fibers. These training programs allowed artificial intelligence systems to gain an increasingly accurate reading of how many fibers were in a given sample. Systems were even able to learn asbestos fiber look-alikes in samples, mimicking the work of human lab technicians. 

The process of creating these successful fiber-counting artificial intelligence systems was aided by the current human nature involved in asbestos testing. Due to the variable nature of human analysts, regulatory agencies have created very specific criteria for fiber counting, such as size requirements and counting procedures. By inputting these specific criteria into artificial intelligence learning systems, computers were quickly able to learn and match the skill of highly seasoned asbestos analysts. Additionally, the reproducible and quick process of computer counting potentially means that artificial systems may soon outpace the traditional ability of analysts to read a given number of samples. 

On its own, artificial intelligence is still a relatively new field, and research on its application to asbestos testing is incredibly limited. Nonetheless, papers such as the one cited above are a testament to the ability of computational science to fix problems, both old and new.

The Future of Asbestos Testing

As it stands, asbestos testing is a field that is not going anywhere. Although asbestos is banned or highly regulated in the U.S (and most of the World), it is a lingering issue that is found in countless buildings and products to this day. As shown through a few groundbreaking studies, artificial intelligence and machine learning are highly advanced systems that can be applied to the very antiquated industry of asbestos testing. Specifically, it has been shown that computer systems excel at learning and perfecting their ability to accurately count asbestos fibers in air samples. With more work and honing, asbestos labs across the country may likely adopt these systems to increase workflows and aid in the issue of removing asbestos. 

Additionally, other forms of asbestos testing, such as bulk testing, also require a lot of human skill and time. Although they may require more advanced analytical techniques, many researchers are hoping to create learning algorithms to teach artificial intelligence systems how to read bulk samples for asbestos identification. 

In any case, while it is rather old, the story of asbestos usage and asbestos testing is not over. Through newly developed computational methods, researchers have shown that even time-tested practices such as microscopy can be streamlined. 

Moving foreward, we will surely have to keep an eye out for other unique developments such as this in the world of computational science! 

Biotational – The Open-Access Hub for Computational Science

Collaborate. Share. Innovate.

Website: https://www.biotational.com

Email: info@biotational.com

LinkedIn: https://www.linkedin.com/company/biotational/

© 2025 Biotational. All Rights Reserved.

This article is published under a Creative Commons CC BY-NC license, allowing for non-commercial sharing with proper attribution.

Want to share your research? Submit your article on Biotational today by emailing info@biotational.com!