Genetic Engineering and Biotechnology News

Future of Autonomous Factories with Advanced Robot Arms and Computer Vision. Digital Twin Simulation of Robotic Automation in High-Tech Manufacturing and Industry 4.0 Applications. Advanced Production

Digital Shadows to Aid Complexity in Handling Process Intensification

Credit: gorodenkoff/ Getty Images Plus

In silico AI-augmented models that report on manufacturing processes rather than controlling them offer a convenient way to help companies engage in process intensification without causing regulatory issues. That’s the view of biotechnology expert William Whitford, PhD.

According to Whitford, who is the founder of Oamaru BioSystems, digital shadows are like digital twins in being real-time digital copies of a process or piece of equipment. Unlike them, however, they aren’t designed to actively control a process. Instead, they simply report back on what’s going on.

This makes them easier to set up than digital twins, he says. “They’re not as comprehensive [as a digital twin], but that might be all you need if you don’t want active control for regulatory reasons.”

Nonetheless, he believes digital shadows are valuable for managing the large, complex datasets being generated by today’s bioprocess analytics. “Machine learning can handle the complexity, the amount, the nonlinearity and high dimensionality of the data we’re getting,” he says.

Such data can include time series, imaging and analytical data, as well as historical data collected during process development.

“For example, you might take the outgassing of a [bio]reactor, look at carbon dioxide and oxygen levels and, from there, infer the metabolic activity,” he says. “And [if you add the historical data] you can infer the number of viable cells.”

By contrast, he says, more simple analytics, such as an equation or simple algorithm, can’t digest the variety of data coming in.

The digital shadow, he says, can make inferences or future predictions by simultaneously integrating process measurements, analytical results, biological observations, historical manufacturing data, and contextual production information. This, in turn, can help manufacturers get more product per square foot of plant, per unit of bioreactor volume or time, he says, and thus aid process intensification.

“Once you’ve defined that goal for your process intensification, you want to know what tools are out there and I want people to know about the tools we have,” he says.