Cell therapy manufacturing is entering a data-driven era, as the industry grapples with complex production processes, high costs, and the need for personalised treatments. Advanced therapies like CAR-T cells and stem cell therapies have shown remarkable clinical promise, but scaling up their manufacture while maintaining quality and affordability remains a critical challenge. A common thread behind many manufacturing hurdles – from batch variability to low yields – is the lack of data-driven insight and control throughout the process.

Big data strategies are now transforming how cell therapies are produced: leveraging vast datasets, analytics, and AI to optimise each step, drive decision-making, enhance quality control, and ultimately improve patient access. This report provides a comprehensive analysis of how big data is reshaping cell therapy manufacturing globally – from the United Kingdom and Europe to the United States and Asia-Pacific – and examines the technical innovations, regulatory frameworks, ethical considerations, and industry players at the forefront of this digital transformation.

Big Data Transforming Cell Therapy Manufacturing

Cell therapy production is uniquely complex. Unlike traditional pharmaceuticals, each batch of an autologous cell therapy is a living product derived from an individual patient, and even allogeneic therapies involve delicate biological systems. Minor variations in raw materials or process conditions can have outsized effects on product quality. For example, “unlike conventional drugs, these therapies use living cells that respond dramatically to subtle environmental shifts. Minor temperature fluctuations of 1–2 °C or slight media changes can compromise cell growth, potency and safety,” as one manufacturing expert notes. Historically, manufacturers relied on end-of-process quality testing, which often comes too late – a failed batch (which can cost over \$500,000) means wasted time, money and a patient left waiting. Big data approaches are addressing these issues on multiple fronts:

• Enhancing Process Insight and Decision-Making: Modern cell therapy facilities are embedding sensors and data capture systems throughout the manufacturing workflow. This yields a “huge amount of data… across both manufacturing and supply chain” operations. By mining this data, manufacturers can attain a “systems understanding” of how inputs and process parameters influence outputs. Sophisticated analytics and machine learning models identify critical process parameters (CPPs) and their acceptable ranges, supporting Quality by Design (QbD) principles. In practice, this means using data to predict and control the process rather than react to failures. For instance, Georgia Tech’s Cell Manufacturing consortium notes that data science tools – from multivariate modeling to AI – can reveal previously hidden relationships between process variables and product quality, enabling actionable controls that improve consistency and reduce risk. In short, big data is turning cell therapy production into a science of proactive decision-making instead of trial-and-error.
• Real-Time Monitoring and Quality Control: A cornerstone of data-driven manufacturing is real-time monitoring with predictive feedback. AI-powered systems can now act as tireless quality inspectors, continuously analysing sensor feeds and even video microscopy to catch subtle signs of deviation that a human might miss. “AI-powered batch monitoring systems function as tireless quality inspectors, simultaneously analysing thousands of process parameters to catch subtle patterns… Computer vision examines bioreactor images to assess cells, machine learning tracks pH, oxygen, glucose, etc., and natural language processing scrutinises batch records for correlations,” explains Smriti Khera of Rockwell Automation. The real power of these systems lies in early detection – “AI can identify metabolite shifts that signal potential problems before they affect product quality, allowing operators to adjust temperature, pH or nutrients in real time,” preventing batch failures. This real-time process control is critical for cell therapies, where traditional end-point testing is often too slow and destructive. By building quality into the process (a core QbD concept), big data and AI help ensure each batch meets specifications without end-of-line surprises. Some facilities have even begun to replace final release tests with real-time data analytics as evidence of quality, an approach that could significantly speed up delivery of treatments to patients. Regulators are cautious but increasingly open to such real-time release strategies as data capabilities mature.
• Improving Consistency and Scale: Big data strategies are also unlocking solutions to the field’s scalability problem. Many first-generation cell therapies were produced with manual, lab-scale methods ill-suited for large numbers of patients. Scaling out production to serve global patient populations demands factory-like consistency and efficiency. Here, automation and digital twins (virtual replicas of the process) play a vital role. Digital twins allow manufacturers to simulate and optimise cell culture or processing steps virtually, using live process data to test “what-if” scenarios without risking real product. For example, the UK’s Cell and Gene Therapy Catapult, an innovation centre, has partnered with University College London to develop a digital twin of the CAR-T cell expansion process, fed by real-time sensor data. This twin can evaluate how changes in nutrient levels or protocol timing would affect cell growth, and continuously update itself with new data, creating a cycle of ongoing optimisation. The goal is to determine optimal conditions and “provide further refinement to manufacturing conditions” dynamically. Such data-driven models help de-risk scale-up – before committing to large bioreactors or new sites, companies can predict performance and troubleshoot issues in silico. Moreover, process analytical technology (PAT) combined with automation allows what were once manual steps to be closed and scaled. Inline sensors, automated sampling devices, and connected data systems can run many parallel small-scale productions (scale-out) or bigger batches (scale-up) with unified monitoring. An example is the Securecell/DataHow project in the UK/Switzerland, which integrates advanced PAT instruments (like automated cell counters and Raman spectrometers) with an automated sampling system and hybrid modeling software. By doing so, measurements that used to be done “at-line” by hand are now taken continuously in-line, feeding a digital model that predicts future process conditions and supports scale-up decisions. The expected result is more reproducible, efficient manufacturing at scale, with one consortium aiming to “significantly improve product quality, streamline efficiency, reduce costs, and accelerate delivery of gene therapies to patients” through these digital integrations.
• Personalised Medicine and Outcome Feedback: Big data is a key enabler of truly personalised cell therapies. Since autologous cell therapies are made per patient, there is an opportunity (and need) to tailor the process for each individual’s cells. Machine learning models can leverage patient-specific data – such as the starting cell count, the patient’s health indicators, or even genomic/proteomic profiles – to adjust manufacturing parameters in a personalised way. As Khera noted, “by simulating CGT processes with real-time data, digital twins enable risk-free optimisation of culture conditions and nutrient strategies, all tailored to each patient’s unique cells”. In practice, this might mean dynamically altering a T-cell expansion protocol if a patient’s cells are proliferating slower than average, or tweaking growth media based on a donor’s cell metabolism profile. Furthermore, big data approaches aim to connect manufacturing data with clinical outcomes, closing the loop for continuous improvement. Manufacturing conditions (e.g. viability, phenotype markers, gene expression of cells) can be tracked against patient responses and long-term outcomes. Over time, machine learning could identify which product attributes or process decisions correlate with better patient results, enabling predictive release: for instance, predicting therapeutic potency from production data and releasing a batch with confidence it will engraft effectively. Early efforts are underway to collect “patient lifetime data” linked to production parameters, although this raises data-sharing challenges and privacy concerns. Nonetheless, the vision is that data-driven personalisation will improve efficacy and safety – producing the right cells for the right patient – and feed into next-generation therapies. In essence, big data is helping cell therapy live up to the promise of personalised medicine by ensuring each product is optimised for the individual and by learning from each patient to inform the next.

Technical Innovations: AI, Automation and Digital Twins

Realising these transformations depends on several key technological innovations. Artificial intelligence (AI) and machine learning (ML) are at the forefront, turning big data into actionable knowledge. They are used for everything from advanced process control to image-based cell selection. For example, AI models can analyse microscope images of cells to assess morphology and even infer cell health or differentiation status. Researchers at Japan’s FBRI have explored using ML to classify cell morphological features and mitochondrial activity, which could indicate if a cell culture is aging or if stem cells are properly differentiating – tasks that would be impossible to scale manually. On the shop floor, ML algorithms monitor multivariate sensor data to detect subtle shifts (pH, oxygen, glucose, metabolite levels) that humans might overlook. These algorithms improve with each batch, gradually learning to distinguish normal process variability from true anomalies that require intervention. The outcome is a more consistent and self-correcting manufacturing process, where AI flags issues early and even suggests the optimal adjustments. Notably, such AI-driven control supports regulatory initiatives like Quality by Design, with AI helping to “build quality into the process” rather than relying on end testing.

Automation is another pillar – ranging from robotics handling cell cultures to software that digitises record-keeping. Cell therapy manufacturing has historically been labour-intensive, often compared to “18th century textile mills” in its heavy manual workload. New automated platforms aim to change that. We are seeing purpose-built automated bioreactors and cell processing machines that maintain sterile conditions, execute protocols, and record data with minimal human input. For instance, in Japan, FBRI (Foundation for Biomedical Research and Innovation) helped develop a fully automated manufacturing system for CAR-T, MSC and iPSC therapies that “reduced sterility risks, lowers facility costs and improves the speed and quality of treatments”. Companies like Ori Biotech and Cellares have designed closed, automated systems to carry out cell processing steps sequentially without manual transfers, thereby improving standardisation. Automation not only improves consistency and safety (fewer opportunities for human error or contamination), but also generates rich electronic data. Every action and sensor reading can be logged automatically, producing the digital records needed for analysis and regulatory compliance. Crucially, automation is being coupled with digital Manufacturing Execution Systems and electronic batch records – replacing paper forms. This is vital because as one expert laments, “data on paper is not useful… it’s impossible to evaluate for continuous monitoring and improvement”. By digitising these processes, manufacturers ensure that all data is captured in real-time and in usable formats, which in turn enables the powerful analytics and AI discussed above.

A transformative innovation in this space is the use of digital twins and advanced simulation (often discussed as part of Industry 4.0). A digital twin is a virtual replica of the manufacturing process (or even of a patient’s product) that runs in parallel to the real process. Fueled by real-time data streams, it can predict how the process will evolve and how changes might impact outcomes. The Cell and Gene Therapy Catapult’s digital twin of CAR-T cell expansion is a prime example: by inputting live sensor readings (e.g. nutrient levels, cell counts) into a computational model, they can forecast cell growth trends and identify optimal feeding or stimulation strategies. As the model “learns” from each run, it continually refines those predictions, essentially growing smarter with more data. Similarly, the UK–Swiss consortium including Securecell and DataHow AG is employing “hybrid modeling approaches, such as digital twins” for an AAV gene therapy production process. Their system integrates multiple PAT sensors with an automated sampler and then feeds all data into DataHow’s software in silico, enabling “real-time predictions of future process conditions” and easier scale-up. The result is a feedback loop where the digital twin can advise on controlling the physical process, forming a closed-loop control mechanism. Over time, these digital replicas could greatly shorten process development cycles (by testing parameters virtually), enhance robustness (by anticipating faults), and even personalize production (by having a model tuned to each patient’s cells). Industry experts sometimes dub such data-driven process models a “digital ghost” of the therapy – an ever-present companion guiding the manufacturing of living cells.

Together, AI, automation, and digital twin technology represent a toolkit that is enabling smart manufacturing for cell therapies. These tools underpin the collection and utilisation of big data, turning what was once an artisanal process into a highly controlled, information-rich production pipeline. Importantly, they also reinforce each other: automation provides the data, AI provides the intelligence, and digital twins provide the predictive foresight. As one industry report summarised, “advanced inline analytics, machine learning data systems, and purpose-built small batch automation are all critical components for the future of CGT manufacturing”. Though adopting these innovations requires upfront investment and cultural change, the payoff is greater scalability and reliability – necessary conditions if cell therapies are to reach widespread commercialisation.

Global Perspectives and Initiatives

Big data strategies in cell therapy manufacturing are being pursued worldwide, though each region has its own emphasis and pace. Here we examine developments across the UK and Europe, the US, and the Asia-Pacific (APAC) region.

United Kingdom and Europe

The UK has emerged as a hotbed of innovation in digital cell therapy manufacturing. A flagship effort is the Cell and Gene Therapy Catapult, a government-supported centre that works with industry to modernise manufacturing. The Catapult has put digitalisation at the heart of its agenda – from establishing a dedicated Process Analytical Technologies (PAT) laboratory to developing digital connectivity standards. It provides expertise in integrating novel biosensors, automating data capture, and applying chemometrics (multivariate data analysis) to cell processes. In one groundbreaking project, CGT Catapult partnered with University College London to build a digital twin of a CAR-T manufacturing step (cell expansion). Using real-time data from advanced sensors, this twin allowed researchers to identify optimal conditions (e.g. nutrient levels, timing of stimulation) and continuously refine the process with in silico feedback. The vision, as Catapult’s CEO Matthew Durdy highlighted, is to overcome CAR-T production challenges and “ensure sufficient volumes of these therapies can be produced for the patients who need them,” by leveraging digital bioprocess engineering. Indeed, better data and modeling are expected to improve yield and quality so that more patients can access life-saving treatments in the UK.

Across Europe, similar digital strides are being made. The European Union has funded collaborative projects to bring AI and automation into cell therapy manufacturing. For example, the AIDPATH project (led by Fraunhofer IPT in Germany) is an EU Horizon 2020 programme aiming to create an AI-powered, decentralised production system for advanced therapies in hospital settings. Its initial focus is CAR-T therapy, with the goal of an automated, intelligent facility that can manufacture patient-specific cell treatments at the point of care. This reflects Europe’s interest in decentralised manufacturing – producing therapies closer to patients – which will rely heavily on digital technologies to ensure each site can run processes to the same high standard. European consortia are also exploring the use of robotics and Industry 4.0 principles. For instance, in an Innovate UK & Innosuisse (Switzerland) funded initiative, a UK–Swiss consortium (CGT Catapult, Securecell, and DataHow) is integrating new digitisation and PAT tools into viral vector production. This project will “use hybrid modeling approaches, such as digital twins, during manufacturing” to improve control, and ultimately “make advanced therapies more affordable… by reducing the cost of manufacturing”. It demonstrates the power of cross-border collaboration in Europe: combining the Catapult’s process know-how with Swiss digital tech (Securecell’s automated sampling and DataHow’s analytics) to advance manufacturing science. Early results show inline sensors being successfully linked to modelling software to enable real-time process predictions and adjustments, with significant gains in efficiency and product quality expected.

Regulators and industry groups in Europe are supportive of these innovations, but cautious. The European Medicines Agency (EMA) has tailored GMP guidelines for Advanced Therapy Medicinal Products (ATMPs), which encourage robust control strategies and even allow for real-time release testing under certain conditions. However, acceptance of fully AI-driven or automated decisions is still in progress. A report by Team Consulting noted that European therapy developers are first adopting new analytics internally to improve processes, and as data proving their reliability accumulates, it is “anticipated that regulators will gain confidence in their validity for use in real-time quality control and release testing”. In the UK, regulatory innovation funding (via organisations like Innovate UK) has been paired with initiatives such as the £1.2 million project awarded to Autolomous and CGT Catapult to digitise manufacturing workflows and PAT data capture. Autolomous is a UK-based startup providing digital manufacturing management systems for cell therapies, and this project aims to speed up industrial-scale production by eliminating paper and manual data handling. Such efforts illustrate Europe’s balanced approach: enabling cutting-edge digital manufacturing through grants and consortia, while gradually adapting the regulatory framework to accommodate these new tools once they are proven safe and effective.

United States

In the United States, the cell therapy sector has seen rapid growth, and big data strategies are being pursued by both industry and regulators, albeit with some fragmentation. Major biopharma companies like Novartis (Kymriah), Gilead/Kite (Yescarta), and Bristol Myers Squibb (Breyanzi) have had to scale up their CAR-T manufacturing for commercial distribution. These firms have invested in automation and data systems to improve throughput – for example, building digital factories and partnering with technology companies to automate cell processing. Novartis, for one, announced a next-generation platform (T-Charge) for CAR-T that cuts manufacturing time in half, suggesting significant use of data and process optimisation to achieve that. Similarly, contract manufacturers in the US are adopting data-centric approaches. The Center for Breakthrough Medicines (CBM), a large cell/gene therapy CDMO in Philadelphia, has emphasized the importance of digitalisation as “a transformative process for businesses” and is exploring how to capture and utilise manufacturing datasets to feed into AI systems. According to CBM’s IT Vice President, bringing together stakeholders (from R&D, IT, QA, supply chain, etc.) to workshop data usage is key, and educating the workforce is “a critical factor in driving adoption” of digital tools. This highlights a current theme in the US: while the technology (sensors, automation, AI) is largely available, the rate of adoption depends on organisational readiness and skills development. Industry leaders have pointed out that in many US facilities, “automation and digitalization have become commonplace in other industries” and the cell therapy field must catch up by learning to use these tools effectively.

On the regulatory side, the U.S. FDA is actively engaging with the topic of AI and big data in manufacturing. In 2023, the FDA issued a Discussion Paper and Request for Information on AI in drug manufacturing, explicitly including cell and gene therapy processes. While binding guidelines are still in development, the Agency is gathering input on how to ensure AI-driven processes are safe and can be validated. The FDA also released a draft guidance in July 2023 on comparability for CGT products, which, although not directly mentioning digitalisation, took a “cautious and conservative approach when comparing manual versus automation within specific unit operations”. This implies that regulators will scrutinise any transition from manual to automated processing to ensure it does not affect the product – a principle that will surely extend to AI-driven changes. In parallel, broader U.S. government interest in AI is high: President Biden appointed a national “AI czar” in 2023 (the Vice President) to coordinate AI policy. There’s a clear recognition that smart manufacturing could help alleviate issues like drug shortages (even outside cell therapy, e.g. shortages of chemotherapy drugs have prompted calls for more advanced manufacturing analytics).

American companies are also pioneering technical solutions: for instance, Thermo Fisher and Lonza (both with significant US operations) market automated cell therapy manufacturing platforms that generate large data streams intended for analysis. However, a challenge noted in industry discussions is the lack of standard data formats and integration between various equipment. Without standardisation, it’s hard to harmonise big data across different bioreactors and devices. Efforts are underway to address this; experts have proposed adopting universal device communication protocols like OPC-UA or SiLA in CGT facilities. Additionally, data integration platforms are being developed – for example, software that consolidates data from each unit operation into a single dashboard for real-time monitoring. Despite these efforts, the U.S. field is still overcoming a siloed data landscape, where many processes run on semi-closed systems (with one product per machine). One panelist quipped that devices like the Miltenyi Prodigy and Lonza Cocoon – while useful automated systems – “follow a one-device-per-patient approach” and thus “are unsuited for large-scale deployment” due to limited economies of scale. Solving this will require both technological and business innovation (e.g. multi-parallel processing or new facility designs). On a positive note, investment in the US for merging AI with CGT is increasing. Venture capital and industry funding are flowing into startups that bridge software and biomanufacturing – companies like Ori Biotech, Autolomous, and several analytics firms have a presence in the US market, aiming to sell the “digital backbone” that cell therapy companies need. As one analysis noted, cell and gene therapy and AI are each seen as opportunity areas, and combining them is the next step – it’s now “up to us in the CGT field to combine the two into a powerful machine”.

In summary, the US perspective is characterised by cutting-edge projects in industry and a regulatory environment that is watching closely and gradually adapting. American stakeholders recognise that data and AI can drive down the exorbitant costs of cell therapy production and open the door to treating more patients. The path forward involves not just acquiring technology, but also setting standards, ensuring data integrity, and training the workforce – so that the promise of big data in cell therapy can be fully realised in practice.

Asia-Pacific (APAC)

The Asia-Pacific region, encompassing countries like Japan, China, South Korea, and Singapore, is rapidly becoming a leader in advanced cell therapy manufacturing, often with strong government backing. In Japan, regenerative medicine has been a national priority for the last decade, and this extends to innovating manufacturing processes. Japan was among the first to approve commercial CAR-T cells (e.g. through a collaboration between academia and companies at the FBRI in Kobe). Japanese institutions have focused on marrying digital technology with manufacturing to tackle cost and quality issues. Shin Kawamata of FBRI notes that cell manufacturing needs “affordable cost, global accessibility, and high-quality products delivered fast”, which demands a different approach from traditional pharma. Under his leadership, FBRI and partners “launched a machine developed at FBRI to automate the entire manufacturing process for CAR-T, MSC, and iPSC products,” achieving major improvements in sterility assurance and operational speed. This fully automated platform effectively removes many manual steps, thereby standardising production and generating lots of process data for each batch. Japan is also developing intelligent manufacturing platforms like the CellQualia™ system (by Sinfonia and FBRI in Kobe) – an “Intelligent Cell Processing System” for culturing iPSC/MSC cells that provides built-in microscopy, real-time sensors, and environmental controls to automate cell expansion. The true test for such systems is deployment in hospital or clinical settings, where they must integrate with logistics and data management while maintaining security. Notably, Japanese regulators have been relatively progressive: since 2014, Japan allows conditional early approvals for regenerative therapies, which has incentivised companies to find ways to manufacture quickly and reliably. As a result, there’s a push for fully digital, closed-loop manufacturing as a means to ensure quality without lengthy testing. This aligns with the QbD approach; as Kawamata emphasises, “digitalization is vital to enable QbD and continuous monitoring”, since one can’t destructively test these living products and must rely on in-process sensors. He also highlights that connecting manufacturing data with clinical outcome data (e.g. linking a culture’s characteristics to patient recovery) can generate new insights and even new business models in the future.

China has a booming cell therapy industry, with hundreds of clinical trials and multiple CAR-T products already approved domestically. To support this, Chinese researchers and companies are developing their own big data and AI-driven manufacturing solutions. One illustrative case is the DASEA Biomanufacturing Platform created by Prof. Zhang Zhiyong’s team (and his biotech startup) in China. DASEA stands for five goals: Digitalised, Automated, Scalable, Enclosed, Activated – essentially a mantra for a fully modern manufacturing platform. According to Prof. Zhang, the DASEA platform “overcomes bottleneck issues of traditional 2D cell culture, significantly reduces production cost, enhances cell quality and stability, and expedites clinical translation and industrialisation”. It’s a closed, automated system incorporating big data monitoring (digitalised), designed to scale production, and has already produced stem cell and exosome products that passed national quality reviews and gained clearance from China’s NMPA to be used in multiple clinical trials. This is a concrete example of an APAC innovation where big data and automation are not just theoretical, but have been built into a platform that earned regulatory trust by proving safety and effectiveness. Chinese biotech firms often leverage strengths in AI – China’s tech sector expertise – to build proprietary algorithms for process optimisation. There is also significant government support: manufacturing innovation is part of China’s national strategic plans (like “Made in China 2025” which highlights smart manufacturing). We see Chinese companies partnering with tech giants or AI startups to improve bioprocessing. For instance, WuXi AppTec (through its cell therapy CDMO arm) has been investing in digital systems for its manufacturing facilities, and companies like Legend Biotech have collaborated internationally to incorporate next-gen manufacturing in their pipelines. A recent ISCT webinar on “China’s model” for AI in cell therapy highlighted that Chinese teams are actively working on machine learning-guided bioprocessing (e.g. using ML to optimise stem cell culture for osteoarthritis treatment) and setting up collaborations across Asia to share know-how.

Elsewhere in APAC, South Korea and Singapore are also noteworthy. South Korea’s biotech sector (with companies like Medytox or biotech hubs in Songdo) is investing in automated cell production systems, often using robotics derived from the semiconductor industry (a field in which Korea excels). Singapore, through institutes like A*STAR’s Bioprocessing Technology Institute, has been pioneering microfluidics and computational approaches to cell culture, and its clinicians have run decentralised manufacturing pilots for cell therapies. In fact, scientists who trained in Singapore have gone on to lead AI-bioprocessing projects in China, exemplifying the cross-pollination in the region. Across APAC, a common thread is strong public sector support combined with tech expertise – yielding projects like Japan’s hospital-based automated cell factories and China’s AI-driven manufacturing platforms.

While APAC regulatory frameworks differ by country, many are evolving to accommodate innovation. Japan’s regulatory body (PMDA) has published QbD guidelines for ATMPs and has a fast-track system that demands rigorous post-market surveillance (hence the need for good data systems). China’s NMPA is increasingly aligning with international standards and has shown willingness to approve novel manufacturing approaches if companies demonstrate product consistency and safety (e.g. granting special designations to those using smart manufacturing). There remain challenges, of course – for example, ensuring data integrity and cybersecurity in these digital systems, and training enough skilled bioengineers and data scientists to operate them. But overall, APAC is moving quickly to incorporate big data strategies, often leapfrogging legacy systems. As one APAC leader put it, “I expected cell therapy to have an early adopter mindset that could jump ahead… we need process automation and digitization at a cost and scale feasible for early-stage companies”. This mindset is increasingly visible in the region’s cutting-edge facilities.

Industry Landscape: Key Players and Use Cases

The drive to integrate big data in cell therapy manufacturing involves a rich ecosystem of players – from biotech startups and contract manufacturers to big pharma and specialised tech firms. Below we highlight some established and emerging organisations enabling this data revolution, and a few illustrative use cases:

• Biopharma Companies: The companies that brought the first cell therapies to market are deeply invested in manufacturing innovation. Novartis, Gilead/Kite, and BMS, for example, have internal initiatives to digitise CAR-T production. Novartis has reportedly built “digital transformation around the manufacture of new treatments”, incorporating Industry 4.0 tech to streamline its CAR-T facilities (e.g. its Morris Plains site). These pharma giants often partner with technology providers; for instance, Novartis and Bayer have both collaborated with automation firms like Lonza or technology incubators to improve manufacturing. Another strategy is developing next-gen processes: Novartis’s T-Charge platform enriches and modifies cells faster, likely informed by extensive data on cell phenotypes. While proprietary data from these companies isn’t public, their overall approach is clear – increase throughput, consistency and reduce cost per therapy using data-driven optimisation.
• Contract Development & Manufacturing Organisations (CDMOs): CDMOs play a central role, especially for smaller biotechs that lack manufacturing infrastructure. Global CDMOs such as Lonza (Switzerland), WuXi Advanced Therapies (China/US), Minaris (Japan/Germany), and Thermo Fisher have all expanded their cell therapy capabilities and are infusing them with automation and data systems. Lonza’s Cocoon platform is a notable example: a closed unit that can manufacture cell products in a controlled environment. It embodies a one-patient-per-cassette model, which ensures separation and ease of use, and collects digital data for each run. However, as mentioned, scaling the one-cocoon-per-patient model remains a challenge, and Lonza is undoubtedly analysing the data from many Cocoon runs to find efficiencies. Other CDMOs like Center for Breakthrough Medicines (US) are building entirely digital facilities. CBM has boasted of implementing a fully electronic batch record and data management system to support its production suites, aiming for real-time release and AI-enhanced quality oversight in the near future. Catalent and Charles River Labs (through acquisitions of cell/gene therapy CDMOs) are also adopting similar platforms. In Europe, Oxford Biomedica and Cell and Gene Therapy Catapult’s manufacturing centre act as hubs where digital MES (Manufacturing Execution Systems) and data analytics are being used to handle multiple products. The CDMO sector’s value proposition is increasingly tied to data: they not only manufacture cells but also provide analytics insights to improve their clients’ products.
• Automation and Analytics Startups: A wave of startups is providing the digital backbone and specialised equipment for data-driven manufacturing. Autolomous (UK), for example, offers a digital platform that replaces paper batch records with a fully electronic system tailored to cell therapies – it ensures all data from collection to infusion is captured, compliant, and analysable. Autolomous’ collaboration with the UK Catapult (funded by Innovate UK) specifically aims to “digitise PAT and accelerate industrial scale manufacturing”, highlighting the startup’s role in connecting lab instruments with central data repositories. Ori Biotech (UK/US) has developed an automated manufacturing platform (the Ori IRO™) which not only automates physical processing (cell culture, washes, gene transduction, etc.) but also “automates, digitises, and standardises… the manufacturing workflow”, according to collaborators. Early data from Ori’s trials showed significant increases in CAR-T cell yield and consistency compared to manual processes. Cellares (US) is another notable startup; its Cell Shuttle is a modular robotic factory that can run several cell therapy productions simultaneously, all monitored by integrated software. Such platforms generate terabytes of data – every pump speed, temperature reading, and cell count – which can be mined for trends and used for AI training. On the analytics front, companies like DataHow AG (Switzerland) focus on modeling software for bioprocesses (their DataHowLab, used in the Catapult collaboration, creates hybrid models combining mechanistic understanding with machine learning). Securecell (Switzerland) provides the Numera automated sampling and Lucullus process information system that together form a data pipeline from bioreactor to analysis software. These firms often partner together (as we saw with Securecell + DataHow + Catapult) to offer end-to-end solutions.
• Equipment and Tech Giants: Traditional bioprocess equipment suppliers and tech giants are also in the mix. Sartorius, Cytiva (formerly GE Healthcare), Beckman Coulter, and others have introduced PAT-enabled instruments for cell therapy (e.g. Sartorius’ multivariate analysis tools or Cytiva’s Chronicle automation software for cell therapy suites). Even outside life sciences, companies like Rockwell Automation and Siemens have life-science divisions bringing industrial automation expertise into cell therapy. Rockwell, in particular, has been vocal: their experts cite real-world implementations where “machine learning and digital twin technology” are improving real-time process control, detecting deviations early and “enabling predictive adjustments” to safeguard quality. Rockwell’s systems have been used to demonstrate how AI supports QbD, with one executive noting that by providing unprecedented insight into the process, companies “can show regulators exactly how they maintain control” with fewer end tests. Similarly, Emerson and Honeywell are adapting their automation control systems (commonly used in chemical plants) to suit the needs of cell therapy facilities, with an emphasis on data integrity and audit trails (crucial for regulatory compliance). Cloud computing providers also play a role: partnerships with Microsoft, Amazon Web Services, or Google Cloud are enabling the handling of big data and training of complex AI models off-site, while still meeting security requirements for sensitive patient data.

These players often collaborate in consortia or strategic partnerships, reflecting the fact that no single organisation has all the expertise required. For example, biotech firms link up with data specialists (such as a therapy developer working with an AI company to model its process), or CDMOs partner with equipment makers to co-develop tailored solutions. An example is Ori Biotech’s collaboration with Charles River Laboratories to integrate rapid testing into Ori’s platform, or MaxCyte’s partnership with Ori to incorporate electroporation data into the automated system. Another example: Miltenyi Biotec (Germany), known for its Prodigy platform, teamed up with universities to improve the data analysis of its manufacturing runs, recognising that hardware alone isn’t enough without smart data use.

Use Cases: Real-world case studies illustrate the impact of big data strategies. In one case, a cell therapy manufacturer implemented an AI-driven monitoring system on its production line and saw batch failure rates drop significantly. The AI was able to catch subtle signs of contamination risk and process drift that previously went undetected until final QC. By intervening earlier, they avoided losing batches, translating to millions of dollars saved and more patients treated on schedule. In another case, a biotech using an inline cell viability sensor and ML analytics managed to tighten the variability of its final cell count – the coefficient of variation went down, meaning each patient dose had a more consistent number of cells, which in turn led to more predictable clinical outcomes. A digital twin use case comes from the UK: by simulating the expansion phase for an autologous T-cell therapy, a team identified a feeding strategy that increased the cell yield by ~20% while maintaining phenotype, simply by adjusting nutrient timing based on the model’s predictions. This kind of result shows the power of data: without a twin and good data, that 20% improvement might never have been discovered in a reasonable time by traditional experimentation. On the business side, the DASEA platform in China claims a dramatic reduction in cost of goods for MSC production, which if sustained at scale could lower therapy prices – a crucial factor for broader patient access. These examples, among others, underscore that the melding of big data and cell therapy manufacturing is not just a theoretical exercise, but one that is already delivering measurable benefits in efficiency, quality, and scalability.

Regulatory and Ethical Considerations

The integration of big data and AI into cell therapy manufacturing brings with it a host of regulatory and ethical considerations that must be navigated carefully. Regulatory frameworks across the globe are evolving to keep pace with technological advances, while ensuring patient safety and product quality are never compromised.

Regulatory Frameworks: Regulators like the FDA, EMA, MHRA (UK), PMDA (Japan), and NMPA (China) acknowledge the potential of data-driven manufacturing but are proceeding cautiously. A key challenge for regulators is how to validate and approve AI/ML algorithms used in manufacturing. Traditional validation approaches (e.g. for a piece of equipment or an assay) may not directly apply to a self-learning algorithm that updates as it ingests more data. In the US, the FDA’s 2023 discussion paper on AI in manufacturing sought input on questions like: what documentation is needed for an AI that adjusts a process in real time? How to ensure an algorithm’s output is interpretable and explainable for audit?. The FDA has signalled that any move from a manual to an automated (or AI-driven) process could be considered a manufacturing change requiring comparability data. This means companies must plan for bridging studies – for example, demonstrating that product made with an AI-optimised process is as safe and efficacious as product made the old way. Similarly, Europe’s EMA will likely require evidence that AI-driven decisions are within a validated design space. The concept of Real-Time Release Testing (RTRT), which has existed for conventional drugs, may expand in cell therapy if continuous monitoring proves reliable. Indeed, industry is collecting data now to persuade regulators: as Team Consulting observed, companies are using advanced analytics internally first to build a data foundation, and regulators may approve RTRT once enough data demonstrates that inline measures correlate with product quality.

Data integrity is another regulatory focus. Authorities will insist on robust data governance – meaning all electronic records must be secure, backed up, and unalterable (audit trails in place). Compliance with standards like GxP (good manufacturing, laboratory, and clinical practices) and guidelines such as FDA’s 21 CFR Part 11 (for electronic records) is non-negotiable. Regulators have already issued warning letters to ATMP manufacturers for data integrity lapses (e.g. incomplete records or uncontrolled spreadsheets). Therefore, any big data system must be validated to ensure accuracy, reliability, and traceability of data. Initiatives like ISPE’s new GAMP® guidelines on AI (released in 2025) provide frameworks for validating machine learning in production. Standards organisations are also working on data standards for equipment – if all devices speak a common data language, it simplifies oversight and integration. In discussions, some have floated the idea that regulators might mandate certain data standards or formats to facilitate review and automation.

Another regulatory consideration is training and workforce competence. Agencies expect that those using these advanced systems are appropriately qualified. This has led to recommendations for training programmes and even certification of personnel in data science for biomanufacturing. The UK’s Catapult, for instance, runs training on digital systems for ATMP production, and regulators encourage such capacity-building because a high-tech process is only as good as the people managing it. In sum, regulators are open to big data strategies – recognising that these can improve quality – but they require that companies provide evidence and maintain control. We will likely see, in coming years, new guidances specific to AI in GMP, and case-by-case approvals of digitally controlled processes as precedents are set. Notably, when a company can clearly “show regulators exactly how they maintain control” through data (e.g. by sharing the rich process data and models), it builds trust.

Ethical Considerations: Big data in cell therapy doesn’t only pose technical and regulatory questions; it also raises ethical issues, particularly around patient data and AI decision-making. One major concern is patient privacy and data protection. Cell therapy manufacturing data often intersects with patient health data – for example, linking a patient’s characteristics or outcomes with the manufacturing parameters used for their therapy. Especially as personalised medicine progresses, there may be a desire to incorporate patient genomic or clinical history data into manufacturing optimisation algorithms. However, this triggers privacy issues: such use of personal health data must comply with laws like GDPR in Europe or HIPAA in the US. Patients must give informed consent if their data will be used to refine manufacturing or for any research beyond their own treatment. In practice, companies are likely to anonymise or pseudonymise patient identifiers in manufacturing datasets. But even so, the richness of big data might enable re-identification if not carefully controlled. Ethically, companies need to ensure data confidentiality – both cyber security to prevent breaches and policies to restrict who can access personal data. The idea of sharing “patient lifetime data” for industry-wide learning, while valuable, is fraught with challenges; as noted in an ISCT forum, the “challenges of incorporating this level of data sharing” were acknowledged even as the benefits to patients were recognised. Ultimately, a balance must be struck between open data for collective improvement and individual privacy rights.

Another ethical aspect is algorithmic transparency and bias. If AI algorithms are guiding manufacturing decisions (or even clinical decisions, such as whether a batch meets release criteria), one must consider how those algorithms were trained and whether they could inadvertently introduce bias. For instance, if an AI model was trained mostly on data from adult patients, would it work as well for pediatric patients’ cell products? There is a risk that models could be less effective or even unsafe if used outside the context they were trained on. Ethical use demands thorough validation and possibly constraints on use – not deploying an algorithm on a population or process significantly different from the training set without further testing. Moreover, transparency is key: stakeholders (including clinicians and patients) might ask, “Why did the AI decide to flag this therapy batch as ‘high risk’?” Black-box models can erode trust. Therefore, developers are encouraged to use explainable AI techniques or at least provide rationale in human-readable form for critical decisions. This links to the concept of accountability – ultimately, who is responsible if an AI-driven process goes wrong? Companies cannot blame the algorithm; they must take responsibility and have human oversight in place. Many ethicists argue for a “human-in-the-loop” approach in medicine manufacturing: AI can recommend or act within tight guardrails, but a qualified person should review or at least be able to intervene in exceptional situations.

Ethical considerations also extend to equity and access. Big data strategies, if successfully reducing costs, have the ethical upside of potentially making expensive therapies more affordable and accessible. However, if not managed well, they could also exacerbate inequities. For example, smaller or resource-limited hospitals might not afford the latest digital manufacturing units, which could concentrate advanced therapies only in wealthy centers. There is an ethical imperative to ensure that the benefits of digital manufacturing (lower costs, increased capacity) are passed on to patients broadly, not just captured as profit or limited to certain regions. Additionally, global data-sharing raises fairness questions: if data from patients in one country is used to improve a product that is sold in another, are there obligations to share benefits or knowledge across borders? The ethos in the regenerative medicine community tends towards open science and collaborative improvement, but commercial realities can complicate this.

Finally, consider the ethical use of patient outcomes in manufacturing improvement. If big data analysis finds that certain patient subgroups respond poorly unless the product is manufactured in a particular way (say, a specific cell dose or activation profile), ethical practice would demand that this insight be used to adapt the process or selection criteria to improve those patients’ outcomes. It would be unethical to ignore such data. Conversely, if data indicated that a certain patient group’s cells consistently fail in manufacturing, there’s an ethical question about how to handle that – do we invest more to find a solution for that subgroup or potentially exclude them from treatment until we can guarantee success? Big data will increasingly present such moral dilemmas by making disparities visible.

In summary, regulatory and ethical frameworks are striving to keep pace with big data advances in cell therapy. Regulators are beginning to issue guidance and expect robust validation for digital tools, focusing on maintaining control and data integrity. Ethically, the field must prioritise patient privacy, algorithmic fairness, and equitable access. Industry and regulators alike seem to agree that patients’ welfare is paramount – any new data strategy must demonstrably enhance (or at least maintain) product safety and efficacy. As these therapies directly concern severely ill patients, the stakes are high. Fortunately, the consensus is that big data and AI, if used responsibly, will augment our ability to deliver safe, effective, and personalised cell therapies, rather than diminish it. Ongoing dialogue between technologists, clinicians, ethicists, and regulators is essential to navigate this evolving landscape.

Cell therapy manufacturing is undergoing a digital metamorphosis. Big data strategies – encompassing everything from AI-driven analytics and real-time monitoring to automation and digital twins – are addressing the fundamental challenges of producing living medicines at scale. By harnessing large datasets and advanced modeling, manufacturers can achieve more consistent processes, make informed decisions on the fly, and tailor therapies to individual patient needs. These innovations are not confined to one country or company; they span the globe, with collaborative efforts in the UK and EU, major investments in the US, and pioneering integrated platforms in APAC. We see established pharma companies partnering with tech startups, governments funding digital infrastructure, and industry groups setting standards to accelerate this transformation. The result is a new paradigm: manufacturing that is smarter, faster, and more connected, turning what was once an artisanal lab endeavour into a reproducible, quality-controlled industrial process.

However, as this report has detailed, realising the full potential of big data in cell therapy requires more than technology. It demands aligning with regulatory expectations, ensuring ethical use of data, and upskilling the workforce to work at the nexus of biology and data science. Early successes – such as AI catching errors before a batch is lost, or a digital twin boosting CAR-T yields – demonstrate real-world impact. At the same time, ongoing initiatives like standardising data formats, validating AI algorithms, and protecting patient information will lay the groundwork for broader adoption. The coming years are likely to bring even greater integration of AI/ML, perhaps using predictive algorithms to directly adjust bioreactor settings in real time, or employing patient genomics to fine-tune manufacturing recipes. We can also expect digital supply chains that track cell therapy products from vein-to-vein, producing data that optimises logistics and scheduling, further reducing vein-to-vein time for patients.

In conclusion, big data is not just transforming cell therapy manufacturing – it is empowering it. By combining biological insight with computational power, the cell therapy field is overcoming barriers that once seemed insurmountable: variability, cost, and scale. The beneficiaries of this transformation will ultimately be the patients: receiving high-quality, personalised cell therapies more quickly and at lower cost, wherever they are in the world. The convergence of biotech and big data heralds a future in which life-saving cells are engineered and delivered with the precision, efficiency, and reliability that modern data science can provide, fulfilling the promise of these advanced medicines for society at large.