Artificial intelligence could revolutionize cell therapy manufacturing, but the industry needs to get better at gathering and handling data according to an expert, who says companies must stop creating silos.
Making a cell therapy is a technically challenging process that combines complicated genetic modification and culturing steps. Each process has variables that must be monitored and controlled.
At present, rather than looking at entire processes, most manufacturers tend to collect and analyze data at each unit operation. While this approach affords control over the processing step in question, it also creates silos that limit data use, says Dalip Sethi, PhD, commercial leader, cell therapy technologies at Terumo Blood and Cell Technologies.
“AI holds significant promise in the cell therapy and biopharma sectors to improve efficiency, reduce waste, cut costs, and enhance outcomes. However, the industry is still in the early stages of determining which manufacturing parameters most impact patient outcomes and how best to optimize them.”
“A key barrier is access to reliable data. The industry’s tendency to silo data complicates AI’s ability to learn from it. While advanced automated technologies are available, adoption has been slow. Early-stage developers often rely on manual processes, limiting digital data capture, and large datasets for model training remain scarce,” he tells GEN.
Rather than investing in AI systems upfront, cell therapy companies that want to embrace the approach and make the best use of process data need to focus on their IT and monitoring systems, Sethi says.
“To implement AI in manufacturing, cell therapy and drug firms need a robust infrastructure focused on comprehensive data capture throughout key processes like cell collection, logistics, and therapy manufacturing.”
Cell culture is a good example of an area that could benefit from AI optimization with the right monitoring and data management infrastructure in place, Sethi says.
“By integrating biosensors for real-time monitoring, researchers can gather data on essential environmental conditions. With more data on metabolic markers like lactate and glucose, automated platforms can eventually predict cell counts and the ideal time for harvesting, creating more efficient and scalable manufacturing systems.”
“However, the key challenge remains collecting enough data to train the computational models and fully optimize the processes.”
Despite these challenges, Sethi is confident the potential benefits of AI will eventually see more developers invest in the data systems required.
“In the rapidly advancing cell and gene therapy field, embracing AI and machine learning is becoming essential for transforming manufacturing processes. Data collection across the entire value chain plays a pivotal role, and leveraging enabling technologies can provide deeper insights into these processes. As automation continues to expand in manufacturing, integrating various platforms and the data they generate becomes increasingly crucial.”
