AI’s Increasing Role in Clinical Trials at Bayer

Artificial intelligence (AI) is reshaping clinical research in the pharmaceutical industry, offering the promise of reducing the staggering costs and lengthy timelines associated with drug development. Despite notable progress in controlled research environments, the path from AI-powered discovery to real-world healthcare impact remains complex and fraught with practical challenges.

The High Stakes of Clinical Trials

Clinical trials represent the most substantial expense in drug development, with some estimates suggesting research and development (R&D) costs can reach $2 billion per drug. The traditional process can take over a decade to bring a new therapy to market. Pharmaceutical giants, like Bayer, are increasingly relying on AI to accelerate the discovery process and uphold rigorous standards, aiming to deliver safe and effective treatments more swiftly.

Beyond Controlled Settings: Real-World Complexities

While AI tools have demonstrated measurable gains in diagnostic accuracy and research efficiency within controlled settings, emerging evidence points to a persistent gap when these innovations move into daily clinical practice. A 2025 review observes that, although AI excels in structured environments, scalability, data diversity, workflow integration, and governance present formidable obstacles outside the lab.

Drug developers and healthcare providers now face the challenge of responsibly converting AI’s technical promise into benefits that genuinely improve patient outcomes in real-world systems.

Insights from Bayer’s Patricio La Rosa

On the 'AI in Business' podcast, Patricio La Rosa—Head of End-to-End Decision Science in Seed Production Innovation at Bayer Crop Science—offered a candid assessment of AI’s evolving role in clinical development. Three core takeaways emerged from the conversation:

1. Establishing Scalable Sensing Modalities

La Rosa notes that the initial infrastructure costs for any sensing technology—whether in R&D or ongoing trials—are considerable. Though AI enables the analysis of greater data volumes than ever, its design must support both robust clinical practice and financial viability. A critical risk, he emphasizes, is developing models that work well in research but carry prohibitive computational costs when deployed widely in healthcare systems. For example, technology like functional MRI (fMRI) may be valuable during development, yet prove too costly for routine patient monitoring. To bridge the research-to-practice gap, sensing modalities must be both scientifically validated and industrialized for practical healthcare use at scale.

2. Navigating Regulatory and Recruitment Challenges

Despite advances in identifying biomarkers and streamlining trial design, regulatory approval processes remain a necessary constant. Even with expedited pathways, pharma companies wrestle with patient recruitment limitations and the inherent timelines of regulatory review. La Rosa argues that the current priority should be leveraging existing technology to its fullest, rather than waiting for regulatory reforms to catch up. He highlights the importance of differentiating between various AI techniques—such as deep learning versus traditional machine learning or statistics—not only for scientific clarity but to build trust and transparency within organizations. Yet, from the patient’s point of view, effectiveness matters most, as they are primarily concerned with outcomes, not underlying algorithms.

3. Cybersecurity and Data Monetization

With the proliferation of sensitive patient data, La Rosa underscores the rising importance of cybersecurity in all R&D workflows. He paints a picture of a continuous "arms race" between advancing analytical capabilities and increasingly sophisticated cyber threats. As AI systems grow more powerful, the industry must keep pace in protecting patient privacy. The podcast also explores the emerging idea of compensating patients for the use of their data in research—a development that could redefine the dynamics of medical innovation. La Rosa envisions a future where patients not only contribute to, but share in the rewards of successful therapeutic development. Yet, this raises new ethical concerns about balancing individual rights with collective benefit in healthcare progress.

European and Global Context

These challenges are particularly salient for European pharmaceutical companies and regulators, who must navigate a landscape shaped by the EU’s stringent data protection laws and growing public demand for responsible AI use. As organizations like Bayer invest in scalable, ethical, and transparent AI solutions, Europe is uniquely positioned to set new standards for balancing innovation, patient consent, and data privacy.

Conclusion

As AI continues to permeate clinical research and healthcare delivery, leaders like Patricio La Rosa are calling for pragmatic, scalable approaches that reconcile technical promise with operational realities. The path forward involves not only technological innovation, but the careful integration of AI into workflows, robust data protection policies, and evolving patient engagement models.

Read the full discussion at Emerj.