Why environmental labs should embrace machine learning and artificial intelligence

Machine learning (ML) and artificial intelligence (AI) may still seem futuristic. You might think your environmental laboratory can’t handle the technologies. Or maybe you don’t think your laboratory can or should be out on the bleeding edge of computing.

You might be surprised then to learn that machine learning harkens back to the 1950s.¹ However, it might not surprise you to learn that environmental, water, and wastewater laboratories lag behind other scientific disciplines in ML/AI adoption.

Know that you need a modern, Cloud-based laboratory information management system or LIMS to serve as your environmental laboratory’s foundation for ML and AI applications. Focus first on implementing applications that establish simple safeguards against human error, and then build up to others that improve your laboratory’s operational productivity.

Adoption by environmental science lags

ML is a branch of AI where the application of statistical algorithms allows the system to learn, thereby allowing the application of decision making through data interpretation. Largely attributable to big-budget commercial investment, ML/AI made faster progress, with more applications, in well-funded laboratory sciences like biochemistry and pharmacology.

Labs testing air, water, wastewater, or soil samples are no less sophisticated and can expect similar returns. Environmental labs that embrace AI will be more than forward-thinking; they will have future-proofed their IT strategies and leaned their business models using modern laboratory management technologies to reduce error, improve productivity, and increase adherence to quality standards.

Applications for ML and AI in environmental labs

There are plenty of places to start with ML in an environmental lab. Think first of mundane, simple checks for common data entry errors and then build up to complex, multivariate calculations that the technology can so adeptly manage, giving you complete confidence in your lab data integrity. ML is perfectly suited to bridge the gap between years of knowledge of analytical relationships and hard-coded checks to determine if specific relationships exist in your samples.

Initially, internal checks stemmed from well-known relationships between various data points, i.e., allowing for flags where Total N appears less than NO₃/NO₂ or ortho-P is > total P. Now, laboratories can reliably use machine learning to check for much more complex relationships denoting potential interferences, like if one compound or element is present at high levels potentially inhibiting the detection of another. These more sophisticated checks result in more accurate data and avoid costly reruns—big wins for your environmental testing facility.

Progressively building AI capabilities, environmental laboratories can automate tasks, not ones that necessarily eliminate human input, rather ones that assist analysts with consistency and validating work quality. Making your analysts more productive can create new opportunities for growth and innovation in your lab’s daily operations.

A modern LIMS is necessary to support your ML/AI applications

Technological innovation has always moved hand-in-hand with laboratory science. No longer flights of fancy, ML and AI can dramatically improve efficiency and quality. That journey begins with modern LIMS software like Clinisys Environmental Laboratory™.

Clinisys Environmental Laboratory enables air, water, and soil testing labs like yours to move forward into Cloud computing, which opens up new resources and capabilities for not only ML and AI, but also lab data analytics. Learn about Clinisys Environmental Laboratory and other Clinisys Laboratory Solutions that are helping laboratories all over the world modernize their laboratory informatics infrastructures to take advantage of the cost savings, process improvements, and business insights that only these ML/AI technologies can bring.

Rockwell Anyoha. “The History of Artificial Intelligence” Harvard EDU Aug 28, 2017, https://sitn.hms.harvard.edu/flash/2017/history-artificial-intelligence