This blog is a summary of a podcast Medable sponsored that occurred on May 7, 2025 and was hosted by Emerj. You can find it here, https://podcast.emerj.com/scaling-ai-for-clinical-trials-with-damion-nero-of-takeda.
Artificial intelligence continues to influence nearly every industry, and life sciences are no exception. In a recent conversation on the AI and Business podcast, Damien Nero, Head of Data Science in US Medical at Takeda Pharmaceuticals, shared his perspective on how AI is changing the clinical trial landscape. With over 15 years of experience applying machine learning and real-world data to drug development, Nero outlined both the progress already being made and the challenges that still stand in the way of broader transformation. His insights highlight how pharmaceutical leaders can think strategically about deploying AI to balance innovation with operational efficiency.
AI is driving progress in discovery and clinical trials
One of the most immediate areas where AI is proving valuable is in the drug discovery phase. By leveraging new data sources such as genomics, patient histories, and social determinants of health, pharmaceutical companies can identify promising drug candidates more quickly. These data-driven methods allow teams to better understand patient populations and uncover potential therapies with higher precision than traditional research methods.
When it comes to clinical trials, the industry is moving toward decentralized models. The COVID-19 pandemic accelerated the need for remote participation, leading to widespread adoption of telehealth, digital consent forms, and electronic checkups. AI tools now help support these new trial models by automating data collection and analysis, reducing administrative burden, and enabling smoother communication with patients. Chatbots, for instance, can interact with patients to collect symptom updates without tying up healthcare providers. While the process remains largely human-driven, automation has become an important efficiency driver, helping trials run faster and at lower cost.
The biggest opportunities lie in areas with robust data
Nero stressed that AI delivers the greatest impact when large volumes of reliable data are available. Oncology is a prime example. With abundant imaging and diagnostic information, AI tools such as computer vision can now screen through hundreds of thousands of tumor images in minutes, providing objective assessments that support diagnosis and treatment decisions. This level of accuracy is particularly important in oncology, where subjective judgments by providers may sometimes reflect optimism rather than objective reality. By grounding decisions in data, AI helps improve forecasting, treatment monitoring, and identification of unmet patient needs.
Rare diseases, on the other hand, remain a challenge. The limited number of patients makes it difficult to collect enough information to train accurate models. While AI can still play a role, its transformative impact will likely be seen first in conditions with larger patient populations. This reflects a broader truth in AI adoption: scale matters, and the technology thrives where data can be aggregated and applied at volume.
Systemic barriers continue to slow widespread adoption
Despite significant progress, Nero cautioned that major obstacles remain. Fragmented site data is one of the largest. The US healthcare system generates inconsistent information due to frequent switching between insurance plans, varying government programs, and gaps created by compliance regulations such as HIPAA. This leaves pharmaceutical companies with an incomplete view of patient histories, making it harder to design and execute trials with confidence.
Patient trust is another challenge. The pandemic created a lingering skepticism toward both healthcare institutions and pharmaceutical companies. Many patients today avoid facilities unless absolutely necessary, which complicates trial participation and treatment adherence. This lack of trust makes decentralized models attractive but also highlights the importance of transparency and clear communication when AI is involved.
On top of these challenges, regulatory environments vary widely. The United States maintains a relatively flexible framework, while the European Union and Asia impose more stringent standards on privacy, compliance, and approval processes. For multinational pharmaceutical firms, this means AI systems designed for one region may not easily transfer to another. Legal, compliance, and regulatory teams often struggle to reach consensus on how much AI can be used, leading to delays and in some cases abandoning AI-driven initiatives altogether.
Outsourcing and internal trust gaps complicate progress
Another barrier to scaling AI is the widespread reliance on third-party vendors. Many pharmaceutical companies outsource algorithm development, which introduces “black box” systems that internal teams cannot fully validate. Without visibility into how models are built, there is reluctance to use AI outputs for anything beyond operational support. Nero argued that for AI to become a true driver of trial outcomes, companies will eventually need to invest in building internal teams with long-term responsibility for development and oversight.
This has proven difficult, as the industry’s track record of maintaining internal data science groups is mixed. In recent years, many pharmaceutical firms cut staff in genomics research, translational science, and related areas. These layoffs undercut the ability to build institutional expertise and reinforce reliance on outsourcing. From an investment perspective, leadership often views data science groups as costly and slow to deliver return on investment. Yet without these teams, companies lack the foundation to build transparent, trustworthy AI systems at scale.
The path forward lies in incremental wins
Given these constraints, Nero emphasized the importance of starting small and focusing on “low-hanging fruit.” Operational efficiencies like automating administrative tasks, improving data processing, and enhancing patient communication can deliver measurable ROI relatively quickly. By demonstrating savings and productivity gains, data science leaders can earn more buy-in from executives and investors.
These incremental wins create the foundation for more ambitious projects. Over time, as leadership becomes more comfortable with AI and competitors begin to take bold steps, the industry will be positioned to embrace larger-scale transformations. The likely tipping point will come when one major player successfully demonstrates a full AI-driven trial model. At that moment, investors across the industry will push for adoption, accelerating the pace of change.
The next three to five years will shape the trajectory
Looking ahead, Nero expects the fastest growth in AI adoption to occur in discovery, where the risk is lower and the potential savings are highest. Virtual simulations and data-driven candidate testing can significantly reduce the time and cost of drug development, making this area particularly attractive for near-term investment. Beyond discovery, decentralized trials will continue to evolve, with the long-term goal of minimizing or eliminating the need for patients to visit trial sites in person.
The transition from reactive to predictive systems represents the ultimate vision. In this future, AI will not only process information but also anticipate risks, flag issues before they arise, and draft reports or submissions proactively. Achieving this will require alignment across regulatory, compliance, and operational teams, along with stronger internal capabilities. While challenges remain, the direction of travel is clear, and the question is increasingly one of timing rather than possibility.
Conclusion
AI is an operational reality that is reshaping discovery and clinical trials today. The greatest progress is happening in areas with abundant data, such as oncology, while rare diseases and fragmented site data present ongoing hurdles. Patient trust, regulatory compliance, and organizational investment strategies will determine how quickly the industry moves beyond efficiency gains to truly transformative outcomes.
For executives, the lesson is to think strategically about sequencing AI adoption. Focus on incremental wins that deliver immediate ROI, build internal capabilities to reduce reliance on black box solutions, and prepare for a future where predictive AI systems redefine the clinical trial process. As Nero suggested, the industry is waiting for a first mover to demonstrate success. When that happens, the race to scale AI in drug development will be on, and those who have laid the groundwork will be best positioned to lead.
.webp)
