Enabling Digitization in Cell & Gene Therapy
TLDR;
Threat & opportunity: Current headwinds pose a threat to innovation but also opens doors for new approaches.
Digitization is crucial: Companies and regulators advocate the vital need for advanced digitization, tools to support data integration are needed.
Avoiding paralysis: Despite financial pressures, the smartest players aren't waiting; they're actively seeking strategic AI/ML solutions (through partnerships if necessary) in high-leverage areas to de-risk investments and gain critical insights.
Smarter, Faster Decisions: In the midst of unpredictability, many are looking for tools to aid more intelligent and quicker decisions.
Participants
Amy DuRoss, Mayo Venture Partner & Board Member, Kosten Digital
Mariano Tribuj, Co-Founder & CTO at Kosten Digital
Shannon Dahl, Chief Scientific Officer & Strategic Board Advisor at Carve Bio
Terri Gaskell, CTO at Rinri Therapeutics
Abhishek Dhawan, Lead Scientist at Omnia Bio
Graham Ball, Chief Scientific Officer at Intelligent OMICS
Raja Sharif, Founder/CEO at ATMPS
Jing Huang, Chief Data & AI Officer at CareDx
Anastasia Bystritskaia, Senior Global Market Intelligence Analyst at Thermo Fisher Scientific
Carlos Martin, CCO (former) and Strategic Advisor at Rocket Pharmaceuticals
Scott Carter, Senior Director Strategic Operations at Discovery Life Sciences
Tom Murphy, CTO and COO, Title21 Solutions
Applications Across the Board
The sheer breadth of applications for digitization and AI spans the entire CGT value chain, transforming how we discover, develop, manufacture, and deliver these therapies.
Terri Gaskell shared her experience with AI in image analysis, noting, "Being ultra niche, with a clearly articulated offering might be better."
Carlos Martin discussed the use of AI in patient identification. "As developers, we have many applications available to us, but the key is the questions that you're looking to answer," he stated.
Abhishek Dhawan highlighted a critical application for complex autologous therapies: "AI can be used to predict cell expansion failures and monitor critical quality attributes, helping reduce costs and improve outcomes."
Graham Ball highlighted the power of digital twins: "Digital twins of patients can help predict responses to therapies and identify reasons for non-responsiveness."
The Power of the Right Question
Connecting disparate data points is opening new possibilities. Scott Carter raised a critical use case: the strategic utilization of patient donor data. Transforming fragmented information into a unified resource, this could involve identifying the "best donors" for particular CGT platforms based on specific attributes like higher NK cell populations, and encompassing logistical factors like donor availability. This integrated approach could revolutionize paper-heavy processes for GMP-level product manufacturing and streamline reporting of donor and product testing results (Certificates of Analysis). Ultimately, a holistic, data-driven approach to donor management could fundamentally accelerate research, optimize product development, and critically, improve patient outcomes.
Integration is No Longer an Afterthought
Data integration is often treated as an afterthought, but it needs to be prioritized. Fragmented systems and siloed data hinder the true potential of digitization.
"For example, launching an ordering platform might help streamline sales. But if it doesn’t connect cleanly with your inventory system, shipping partner, and finance software, you’re still left stitching everything together manually. Integration, in this context, isn’t a technical step. It’s what allows the entire operation to move faster, smarter, and with fewer errors."
Tom Murphy articulated the challenge from a data perspective: "The data is out there, but these datasets are not designed for interoperability. They're built for the operational use, not for broader analytical purposes."
Jing Huang further emphasized data quality: "One of the first things I came here to do is integrate data, and we have made strides in that area. The quality of the data and the sheer amount of the data really decides the quality of your model and what type of model you can build."
"One of the first things I came here to do is integrate data, and we have made strides in that area. The quality of the data and the sheer amount of the data really decides the quality of your model and what type of model you can build."
Amy DuRoss highlighted a critical challenge: "Multiple existing platforms create complexity for end users. A unified, user-friendly interface with backend integration via APIs is preferred for efficient adoption." Her insight stems from addressing "paper-heavy" logistical and data management hurdles in CGT supply chains, providing "smart plumbing" for this intricate ecosystem.
Graham Ball highlighted a powerful solution for data privacy in collaborative research: "Using cloud compute and federated models can allow analysis without sharing proprietary data, solving privacy concerns." This is a game-changer for the CGT sector, where proprietary data is a goldmine, and privacy and security are paramount.
Surviving vs. Thriving – Smart Solutions for Broader Access
The fundraising landscape for CGT is a gauntlet. Many organizations are in "survival mode," making it hard to justify significant digitization investments. How can we offer smart solutions enabling broader access and adoption, even with limited budgets?
As Amy DuRoss pointed out, "Many companies are in survival mode with limited budgets for Al and digitization. Cost-effective scalable solutions and clear ROI are necessary to drive investment."
Amy further elaborated on how AI can directly impact accessibility: "AI can be used to create predictive models for payers, helping them understand outcomes and improve accessibility for patients."
One participant emphasized the urgency: "Priority number one, two and three are things we solve in the next six months or we're out of business." This means solutions must be impactful and quick.
"This is a field that, outside of Big Pharma, has had to be more scrappy and resourceful than ever."
As one participant noted, "This is a field that, outside of Big Pharma, has had to be more scrappy and resourceful than ever." Focusing on urgent, high-impact problems, even with limited resources, can yield disproportionate returns.
Prepare for scale through early integration
True progress in digitization hinges on effective application integration and customizing solutions for specific needs.
"None of this means integration is simple. It takes planning, especially when working with legacy systems or third-party platforms that don’t expose clean APIs. But the alternative - building disconnected tools and retrofitting them later - is almost always more expensive and more painful."
What does a digital solution designed with integration in mind look like?
Building with open standards: Ensuring new tools can "talk" to existing systems.
Shared data models: Creating a common language for data across different applications such as standardizing patient eligibility criteria across clinical trials, harmonizing manufacturing process parameters for consistent batch records, or unifying adverse event reporting formats for pharmacovigilance.
Investing in integration middleware: Tools that act as translators between disparate systems such as Qlik, Mulesoft and InterSystems.
This iterative approach allows organizations to start small, see immediate value, and scale their digitization efforts. It's about building practical, adaptable solutions for each clinical development phase. Ultimately, true digital transformation isn't about grand overhauls, but precisely tailored solutions addressing specific, urgent pain points.
What's Next?
The conversation doesn't end here. Insights from this roundtable are just the beginning. Sharing targeted use cases and reflections will help the CGT community navigate digital complexities.
As Carlos Martin wisely stated, "Collaboration with regulators and stakeholders is essential to educate and align on AI's role in the industry." It requires open dialogue, shared learning, and collective problem-solving across the entire ecosystem—from innovators and developers to regulatory bodies and patient advocacy groups.
Graham Ball also observed that "Large language models specific to FDA submissions could benchmark success criteria and streamline the regulatory process." This highlights a powerful future for AI in accelerating therapies. Imagine LLMs trained on FDA submissions, identifying patterns, predicting roadblocks, and drafting compliant documentation. This augments human expertise, allowing regulatory teams to focus on high-level strategic review.
Anastasia Bystritskaia further underscored the global imperative for AI adoption, stating, "Regulatory alignment across global regions is needed to speed up AI adoption and avoid regional disparities." For CGT to scale globally, a harmonized approach to AI regulation is critical. Avoiding a patchwork of regional rules prevents bottlenecks, ensuring efficient deployment of digital solutions across borders.
We'll be continuing this series, diving deeper into specific areas like data governance, regulatory pathways, and the ethical considerations of AI in CGT. How can we ensure these vital discussions and initiatives are better linked up to create a cohesive and accelerated path forward for the entire CGT ecosystem? Stay tuned for more opportunities to engage!
