Scientists develop remarkable tool in fight against widespread threat to human health: 'It's important for the people making those calls'

The more we learn about microplastics pollution, the more we've come to realize just how troublesome it really is, but a group of scientists recently developed a technology that could help alleviate the problem by improving our knowledge of it.

According to Phys.org, researchers in Michigan have developed a way to better identify the chemical makeup of microplastics collected from the environment.

Microplastics are everywhere … literally. They're in the oceans, our waterways, the air, the soil, the food we eat, and our bodies. Research has shown that microplastics are connected to a plethora of severe health complications, including cancer, developmental issues, autoimmune conditions, and neurological diseases.

Scientists determine what chemicals make up a sample of microplastic using spectroscopy, which entails observing how a substance reacts to light. It either absorbs or scatters light and then creates a unique pattern called the spectrum, which identifies the substance's "chemical fingerprint."

The current process involves researchers feeding a dataset of spectra whose identities are known to train a machine learning algorithm that learns to predict a substance's chemical makeup. The problem is that some plastics have similar chemical makeups, which can lead to similar chemical fingerprints, leading to uncertainty in the identification process.

While the newly developed process, using conformal prediction, doesn't eliminate uncertainty, it allows researchers to at least narrow down the microplastic's chemical makeup. To do this, researchers train the machine learning algorithm with a calibration set of known spectra. The conformation prediction then analyzes the differences between the predictions and the correct identities, and it adds other possibilities.

Watch now: Giant snails invading New York City?

Rather than predict a single identity with uncertainty, the new process gives a set of possible identities with a high level of confidence that one of them is correct.

The team brought in an expert to manually label the particles with the correct identity, and every set produced by the new process contained the correct identity.

Developments like this will help researchers better understand the growing problem of microplastics pollution by giving those researchers a better idea of what chemical makeups are being found where. This could help inform mitigation and cleanup efforts, as well as future legislation tackling the issue.

Ambuj Tewari, a University of Michigan professor who took part in the research, wrote in The Conversation: "Data from microplastics analyses can inform health recommendations and policy decisions, so it's important for the people making those calls to know how reliable the analysis is."