How AI Is Poised to Transform the Life Science Lab

AI is a buzzword. Ever since the release of the generative AI app ChatGPT—and immediately followed by several others—AI seems to be everywhere. Life sciences laboratories are no exception. From helping to understand vast datasets and generate new insights from existing data, to generating data itself, AI can play a role. It is even seen in more seemingly mundane tasks like monitoring and optimizing lab equipment, and assuring that equipment and the data generated by it, as well as the networks that carry it, are secure.

What is AI?

Artificial intelligence (AI) is a loose term. For the purposes of this article, AI is synonymous with the ability of a computer to mimic human thought processes and use the results to perform complex tasks. It can be taught to categorize patterns and correlations, and to detect anomalies. After the appropriate training it will be able to write an essay answering a question; to identify a given person in a photo of a crowd (finding the proverbial needle in a haystack); to decide whether a patient has cancer based on images and tests.

AI helps with the science

One of the things AI excels at is its sheer ability to process far more data than a human can, and to find patterns that would evade even the most adept human.

For example, there are hundreds of thousands of antibody sequences found in public and private databases. An effectively off-the-shelf large language [a type of AI] model can be trained on these sequences, learning what they look like and some of the properties associated with them, says Philip Kim, Professor in the Departments of Molecular Genetics and Computer Science at the University of Toronto. These might include avidity, solubility, and immunogenicity, among others. “You can have these models suggest mutations to existing antibodies, and more often than not these suggestions lead to an improvement of antibody properties. You can also use the models to generate new antibodies.”

AI can be used to integrate data from sources such as genomics, proteomics, existing therapies, as well as clinical trials. In this way previously unrecognized connections may further research on current biopharma projects, for example, or suggest ways to repurpose existing drugs.

Another example is the ability of AI to use natural language processing to extract relevant findings from the published literature. This is enabled, in part, by AI’s ability to understand many different ways of relating findings that a heuristic algorithm—no matter how sophisticated—would be unable to do.

Along similar lines, companies have created AI tools that can extract relevant data from electronic lab notebooks (ELNs) and lab information management systems (LIMS) across a company. It can then use AI to analyze the data flows and find correlations and properties that may have been discovered but missed by the individual scientists.

Because of its ability to understand natural language, AI allows researchers with less computer expertise than was previously required to perform some complex analyses. AI can even power a virtual assistant that will answer questions, schedule experiments, help analyze and interpret data, as well as suggest further avenues of research.

AI helps run the lab

“Researchers have made significant strides in life science labs by incorporating AI into lab instruments and informatics platforms,” explains Sergey Vlasenko, Associate Vice President, Pharma and Biopharma End Markets, Agilent Technologies. “Lab instrument intelligence—the fusion of advanced technologies and algorithms with lab equipment—is a part of this transformation.” Yet such intelligent instruments go beyond automation—they can self-monitor, self-diagnose, and even self-optimize.

He cites two examples: “High-performance liquid chromatography (HPLC) systems can dynamically adjust flow rates, column temperatures, and solvent compositions based on real-time data and changing conditions. Another example is advanced AI software that automates the labor-intensive task of gas chromatography/mass spectrometry (GC/MS) data analysis. This streamlines the process and improves the lab workflow, from sample analysis to reporting.”

In a similar way, labs can be automated end-to-end, with each step feeding into the next without human intervention. AI can learn from the results of each step, feeding back into the system, identifying patterns and informing the processes going forward. Information can be used optimize entire workflows, manage lab—or even enterprise-wide—resources, ascertain the health of the system, and recommend predictive maintenance.

“There’s a lot of AI being used to understand how to optimize the environment, how to optimize the equipment usage, how to optimize workflows,” says Matt Burtch, CEO of Sigsense Technologies. His company, for example, uses AI to improve lab efficiency, examining which pieces of equipment are being used and how, by tracking the power usage signatures—the “voice of the machine”—of each. In the hands of AI this seemingly innocuous readout can indicate, for example, what methods are being used, equipment downtime, whether there is a problem, and when the signature is beginning to deviate (and what that is likely to mean). Signatures are compared against trends seen in millions of machine years’-worth of data. And “we’re able to address those sensitive scientific environments and patient-facing environments because we don’t have access to any of the scientific data, we’re just collecting power-draw information.”

The future of AI

Ethical concerns such as those around privacy and the potential for disinformation, and concerns about the black box nature of its algorithms, still dampen enthusiasm for a more universal adoption of AI. But there is clearly no putting the genie back in the bottle. AI is here to stay.

“Looking ahead, the broader adoption and integration of generative AI (GAI) will enable unprecedented progress, revolutionizing various aspects of life sciences,” predicts Vlasenko. “GAI's ability to process and analyze vast datasets, generating novel insights, will render it an indispensable tool for life science laboratories.”