Key takeaways: Although banned in March 2024, asbestos still poses serious health risks to the American public. Currently, there’s no approved way to test for asbestos without sending a
sample to a lab for testing. Scientists in Australia and Spain have developed micro and macro-level ways of testing for asbestos by using software trained on prior images and pattern recognition. Hopefully, this technology will be adopted by the United States in the coming years, as efforts to dispose of and replace asbestos become more urgent. AI and asbestos could be start to be used in the same sentence.
Where does Asbestos Still Exist?
Asbestos is present in about 20% of the United States’ water pipes, 30 million American homes, and in consumer products, like makeup and crayons. Even though a finalized asbestos ban was passed in March 2024, Americans will be feeling the residual effects of unchecked asbestos use for years to come.
For Americans that have or suspect asbestos in their home, there are only a few courses of action: safely seal it, so that no fibers can become airborne, or have it abated (safely removed). The key here is not removing or sealing the suspected areas yourself, although there isn’t a way to qualify the presence of asbestos. If your house was constructed between 1930-1950, then there’s a chance of asbestos in your insulation, paint, flooring, siding, or even fireplace. But again, only laboratory testing can ensure that the fibers are asbestos fibers, and not another, similarly presenting material. Obviously, the need for laboratory-grade testing has complicated the abatement, removal, and treatment of asbestos, which has made asbestos a more temporally extended issue. In some European countries, there have been efforts to completely remove asbestos—even sealed asbestos—by 2032 (for private buildings). Identifying asbestos in the contexts of bans, removal efforts, and public health is becoming an urgent goal.
How can AI Address the Need for Lab Testing?
Researchers in Australia trained an image recognition software to identify asbestos with approximately 90% accuracy. The software was trained on 7,328 pictures of asbestos in about 1,000 different contexts, so that the software would be able to ID asbestos regardless of location, presentation, or fiber type. The current prototype requires an on-site microscope: a cell phone’s camera lens can scan the magnified sample via the detection software. An answer can then be established on the spot—no laboratory testing or waiting. 90% accuracy is significant; however, as the software learns through more images (and transfer learning, which is the learning of new tasks with the application of similar skills), then the accuracy is bound to increase.
As previously mentioned, Spain is aiming to eradicate all asbestos by 2032. Although an ambitious goal, scientists are training AI learning models—like Australia’s software—to detect asbestos on a macro scale. They’re training learning models to not only identify asbestos, but to anticipate where it might be based on construction history, patterns of usage, and physical presentation. Instead of using a microscope, they’re using satellite-scale images that are within the public domain. This is a massive win for Spain, who, despite banning asbestos, is still dealing with asbestos-related public health crises. Ideally, this kind of technology can be adopted by the United States, especially as more efforts to replace or remove asbestos begin.
If you or a loved one has been diagnosed with mesothelioma, you may be entitled to financial compensation. The experienced team at The Halpern Law Firm is dedicated to helping victims of asbestos exposure get their just compensation. For a free consultation, call (800) 505-6000 or fill out our contact form.
Sources:
https://www.uoc.edu/en/news/2022/174-asbestos
https://www.sciencedirect.com/science/article/abs/pii/S0304389423021374
https://www.mdpi.com/2071-1050/15/5/4276
