Cell therapy, one of the most promising frontiers in modern medicine, holds the potential to cure once-incurable diseases such as cancer and genetic disorders. But for all its promise, manufacturing these therapies remains unpredictable and expensive, with high failure rates and few real-time solutions for process control.

It’s kind of like the wild west meets the gold rush. You have incredibly powerful therapies being developed, but the systems are so complex, and the tools to measure and control them in real time just haven’t kept pace.

When I first joined the U.S. Naval Research Laboratory, I was part of a team that developed machine learning and predictive algorithms for defense applications, including bomb detection, chemical agent sensing, and shipboard fire monitoring. That same multivariate analytical foundation underlies LumaCyte, helping biomanufacturers tackle one of the most complex challenges in medicine: making cell therapy more consistent, more scalable, and more accessible.

Naval defense to biotech breakthrough

The early work in defense sharpened a focus on precision, automation, and multivariate statistical analysis, easing the transition to biotechnology. Before starting the company, we were using lasers to move particles based on their intrinsic properties. We realized that same technique could be used to analyze cells, without labels or staining, based purely on how they behave under optical and fluidic forces.

That approach became the basis for Laser Force Cytology™, or LFC, the technology underlying LumaCyte’s single-cell analysis platform. It provides a high-content, label-free analysis of individual cells in just minutes, allowing scientists to understand what is happening in their cultures far faster than traditional methods like flow cytometry.

Instead of waiting hours or days for results, we can deliver insights that enable our customers to see what’s happening inside a bioreactor in minutes. The resulting speed and process clarity don’t just improve decision-making, it has the potential to fundamentally change how cell therapies are developed, scaled, and delivered.

Real-time cellular insights fast

Cell therapies rely on engineered living cells, which are far more complex than chemical compounds or proteins. Measuring their health, potency, and behavior in real time has long been a bottleneck in manufacturing. A typical assay can take several hours or even days. By the time you get the results, it’s too late to adjust the process.

What we’ve built allows for five-minute insight into what’s happening across a bioprocessing step or manufacturing run. Rapid, label-free analysis captures a wide range of cellular characteristics, including viability/vitality, activation, transduction efficiency, differentiation stage, and potency. More importantly, it can capture these insights early and often throughout the manufacturing process, with little sample volume and processing requirements.

ML for predictive manufacturing

Machine learning is what makes this technology so powerful. By collecting dozens of variables from each cell, our team applies multivariate analysis to identify patterns invisible to the human eye with univariate metrics. With multivariate machine learning, we can pull out information that would be impossible to see otherwise. It allows us to predict, based on a sample, whether a batch will succeed or fail, or even which donor cells will produce the most effective therapy.

In a pre-LumaCyte publication and patent that describe the use of multivariate statistics for spectroscopic chemical analysis, eight chemical components were identified in a mixture only containing six, uncovering hidden molecular species never before observed under neutral pH conditions. A similar multivariate mathematical approach now drives the company’s predictive analytics for live cells. The ability to forecast success is a significant advantage in a field where the cost of failure is high and the timeline is tight.

Whether it’s screening donor cells, assessing the viability/vitality of frozen samples at thaw, or monitoring manufacturing cultures mid-process, the insights generated help ensure that only the best starting materials are used and only the strongest batches proceed.

Adaptive, AI-driven biomanufacturing

The next step is prescriptive analytics. By combining real-time sensing with machine learning and AI analytics, a platform is being built capable of recommending corrective actions before a failure occurs. The future is about adaptive controls. Prescriptive AI doesn’t just tell you something’s wrong; it predicts what will happen ahead of time, and then advises you what to do about it. Eventually, we’ll have autonomous or agentic systems that can make those adjustments on their own or with minimal human supervision.

This level of intelligence is essential for modern manufacturing, as seen in an analogy of self-driving cars. If your Tesla only had a GPS map but no sensors, it might stay on the road, but would drive through red lights and hit obstacles in its path. Automation without analytics is useless in dynamic environments like cell therapy manufacturing. By creating both the sensors and the intelligence to interpret them, LumaCyte bridges one of the most critical gaps in the field.

LumaCyte’s approach has caught the attention of regulators. The FDA invited the company to train its cell and gene therapy reviewers. The agency is pushing for adaptive manufacturing, and they’re expecting companies to prove they can consistently produce safe and effective products, which means tight tolerances, predictive control, and evidence that you understand your process and your product.

Regulatory expectations are rising as more therapies move toward commercialization. Consistency, transparency, and control are no longer aspirational; they’re essential. In this landscape, companies equipped with precision analytics and real-time insight won’t just meet the bar; they’ll set it.

Sean Hart, PhD, co-founder, CEO and CSO of LumaCyte.