The progress in laboratory automation could be one of the most consistent changes in research in the life sciences and biomedicine over the past 30 years. The evolution of this field borders on the miraculous. That perspective, though, comes from someone who is old enough to remember pipetting with a rubber bulb and analyzing data with a calculator—laboratory tasks that used to be common but are seldom performed by today’s scientists. To get a modern angle on the advances in laboratory automation, GEN reached out to a group of experts.

These experts describe a range of challenges that drive today’s laboratory automation landscape and how they are being solved. Not surprisingly, a rubber bulb is not part of any of today’s solutions. Instead, companies create systems of automation, often supported by artificial intelligence (AI).

Automation in an ecosystem

“It’s key to understand that automation can’t function in isolation,” says Carola Schmidt, general manager of automated robotic systems, Revvity. “True automation in laboratory workflows requires seamless integration of the relevant instrumentation technology and the associated software, reagents, and consumables. Integrated automation that connects multiple steps in the laboratory’s workflow enables a reduction in manual workload and frees up time for the scientists to focus on research and innovation.”

Collaboration is central to automating laboratory processes effectively. “The concept of collaboration goes beyond merely offering automated services and tools,” Schmidt explains. “It also requires growing with the evolution of a company or laboratory to better meet its needs. For instance, we acquired Oxford Immunotec, which is now rebranded into Revvity and plays a global role in fighting tuberculosis.” With that acquisition and Revvity’s expertise in automation, the company, Schmidt says, is “able to provide customers with fully automated workflows for tuberculosis testing.” She adds that such workflows could bring advanced tuberculosis testing to more people around the world.

Other diagnostic technologies also move ahead with automation. For example, the UNIQO 160 from Revvity’s EUROIMMUN business fully automates indirect immunofluorescence diagnostics. “It’s traceable with barcodes and can process up to 160 primary samples and 18 slides,” Schmidt asserts. “Moreover, it can be processed in one run on the compact benchtop device, which generates high-quality images and enables reliable AI-based results.”

In fact, AI can improve laboratory automation in many ways. As examples, Schmidt mentions rule-based decision making, pattern recognition, and image analysis. “In its more advanced forms, AI capabilities built into laboratory software and instrumentation can aid in sample handling and other manual processes carried out by robotic systems,” she says. “AI can offer guidance, propose a workflow, and reduce errors, but at the end of the day, a human determines the outcome.”

In many ways, liquid handling makes up the heart of laboratory automation. Last summer, Revvity launched the Fontus Automated Liquid Handling Workstation. According to Schmidt, this platform is “suited for a range of liquid-handling applications and will be especially valuable in this climate where the supply of talented laboratory personnel is limited—even though more work needs to be accomplished in a timely manner.”

Addressing underutilization

All laboratories hope to make the most of their assets, but doing so is challenging. Turnovers and shortages in personnel create key challenges.

“Some laboratories lack the trained personnel required to operate their systems, leading to underutilization of automation solutions,” says Hal Wehrenberg, head of product management, Tecan. “Usability directly impacts the ability to quickly train teams and redeploy resources.” Consequently, Wehrenberg points out that usability is crucial to clearing the underutilization hurdle.

Tecan continually enhances its Fluent Automation Workstation with more flexible features, such as the improved Air FCA (Flexible Channel Arm) with MultiSense technology, which provides AI-based pressure monitor pipetting (PMP-AI). “PMP-AI uses neural networks to interpret pressure curves, helping to identify pipetting anomalies and mitigate the risk presented with challenging samples like whole blood,” Wehrenberg explains. In addition, this platform’s Phase Separator “is a unique, pressure-based capability of the pipetting arm, which works across all eight pipetting channels in parallel, independently detecting the interfaces between liquid layers while aspirating,” Wehrenberg notes. “This is an ideal solution for workflows like plasma separation from centrifuged blood samples.”

Tecan used machine learning in the development of the new Air FCA. For example, “the Phase Separator technology uses advanced data analysis techniques,” Wehrenberg says. “By continually monitoring pressure measurements during aspiration, it is possible to automate the identification and the separation of liquid fractions based on their different viscosities.”

Tecan’s Fluent platforms also include DeckCheck, which Wehrenberg describes as “a camera-enabled feature that helps avoid errors that could impact a run.” To do this, he says, “it displays discrepancies, such as missing plates or incorrectly loaded tip boxes, allowing the operator to correct the error, avoiding the risk of failed runs.”

The company also applied machine learning to DeckCheck. “The goal was simple: improve the end user experience and eliminate failed runs,” Wehrenberg declares. “This required large amounts of data, including data in the form of images of our Fluent systems with various segments, labware, and runners in a wide range of lighting conditions.”

Consequently, the system “can compare a live image with a reference image, identify discrepancies, and alert the user to act,” Wehrenberg says. “This live image analysis feature would not have been possible without the advancements and proper implementation of AI in our product development program.”

Seeking sustainable solutions

“One of today’s biggest challenges in advancing laboratory automation revolves around integrating sustainable practices and reducing environmental impact while maintaining efficiency and accuracy,” says Thomas Deutschmann, vice president of product development life sciences instruments at QIAGEN. “Traditional laboratories often face issues like wastage of reagents, single-use plastics, and high water and energy consumption.”

QIAGEN’s QIAcuity system
In November, QIAGEN launched its QIAcuity Digital PCR kits. These kits can be used in QIAGEN’s QIAcuity systems, which can be applied in applications from cell and gene therapy production to food safety assurance. For example, the QIAcuity Mycoplasma Quant kit detects contaminants in the production of cell and gene therapies.

QIAGEN takes on the sustainability challenges in several ways. For example, QIAGEN has developed a portfolio of purification kits called QIAwave. “These kits,” Deutschmann says, “reduce the use of plastics by up to 62% and cardboard by up to 58% compared to QIAGEN standard kits.” Nonetheless, all of these kits use the same chemistry and provide the same performance. The QIAwave kits, though, consist of fewer components. Waste tubes are made from 100% recycled plastic, and bottles containing suitable buffer concentrates are smaller. Deutschmann notes, “With more compact kits and innovative packaging methods, we can reduce cardboard consumption.” Plus, these kits can be automated.

On the analytical side, Deutschmann mentions the challenges of automating image analysis associated with digital PCR: “QIAGEN’s QIAcuity Digital PCR system serves as an example of the complexity that image analysis must manage within this technology. On a single microplate, the system can process up to 96 wells, each containing 8,500 partitions.” That’s a lot of imaging to analyze. Accordingly, Deutschmann observes, “Ensuring accurate and efficient image analysis is essential, given the minute scale and complexity of the samples.”

To take on complex imagery, QIAGEN applied advanced algorithms. “As we strive to continuously improve image analysis, one approach involves training neural networks to differentiate between valid and invalid partitions,” Deutschmann emphasizes. “This allows for accurate analysis of photographic images of the filled nanoplates, ensuring that all partitions are correctly identified and that potential artifacts, such as dust, are excluded.”

Expanding access and integration

“Despite the many modern advancements made in laboratory automation, affordable automation is still inaccessible for many laboratories,” says Susan Magdaleno, PhD, director of research and translational R&D, sample preparation technologies, Thermo Fisher Scientific. “The low-cost sample prep instruments currently available on the market do not offer a fully automated user experience.”

So, Thermo Fisher developed advances in its economical liquid handling platforms that “allow for low-throughput, walkaway automation from sample to quantitation to assay setup to help streamline workflows,” Magdaleno says. “By automating individual sample handling wherever possible, this reduces user error and removes the tedious workflow steps for the user.”

In addition, Magdaleno notes that automation-related information could be better integrated with laboratory information management systems and electronic laboratory notebook systems. Here, she says, Thermo Fisher is working on “fully integrating sample preparation and data capturing of lot number, quantitation, method, etc.”

“Once information for a sample is entered, it is assigned a barcode and tracked from entering the laboratory through sample prep to storage and downstream applications,” Magdaleno says. “In a perfect world, all instruments would update each sample’s data files with its data/results.”

Magdaleno expects AI to play an increasing role in automation ahead: “For example, AI could come into play to help interpret video images—like how we use the human eye to ensure we are delivering liquid to a tube with a manual pipette.” AI can help ensure that liquids are dispensed in the right volumes where needed.

Magdaleno sees AI coming into automation at the earliest stages of analytical processes. “The integration of AI and machine learning into sample prep instruments could really improve the instrument’s ability to detect sample type, optimize extraction protocols, and troubleshoot errors in real time,” she says. “This could improve the precision and reliability of the process.”

Today and tomorrow

As shown here, laboratory automation continues to provide a range of advances, from accuracy and asset utilization to productivity and sustainability. As Schmidt puts it: “Without question, automation and data management are making room for innovations, bringing lifesaving therapies to people at a rapid rate without compromising on quality or accuracy. We are only just scratching the surface.”

Previous articleLaboratory Automation Reaches Every Stage of Drug Development
Next articleShape-Shifting Protein Study Could Lead to Development of New Drugs