2025 in Review: The Top 5 Trends That Shaped Clinical Research

If there was a single regulatory theme of 2025, it was real-world evidence (RWE). The FDA's December guidance on de-identified data marked a turning point, explicitly allowing sponsors to submit RWE analyses without transferring individual patient records in many use cases.

Key Developments:

• FDA De-Identification Guidance: This landmark guidance removed a major logistical barrier to RWE use, clarifying that properly de-identified aggregate data is acceptable for many regulatory purposes.
• European Health Data Space (EHDS): The EHDS regulation, which came into force in early 2025, established a framework for cross-border use of health data in the EU, opening new opportunities for pan-European RWE studies.
• Successful Approvals: Several drugs and devices received approvals or label expansions supported primarily by RWE, demonstrating regulators' growing comfort with this evidence type.

Looking Ahead:

In 2026, we expect RWE to become a routine component of regulatory submissions, not just an alternative pathway. Sponsors who have invested in RWE infrastructure and capabilities will have a significant competitive advantage.

Related: FDA's New RWE Guidance: How De-Identified Data is Reshaping Submissions

2025 was the year AI in clinical research transitioned from a buzzword to a tangible set of tools delivering real value. Across the clinical trial lifecycle—from drug discovery to patient recruitment to data analysis—AI applications matured and gained adoption.

Key Developments:

• AI Agents in Recruitment: Tools like Grace from Grove AI demonstrated that AI-powered patient relationship management can significantly improve recruitment speed and retention rates.
• BRIDGE Trials: The novel BRIDGE trial design showed how AI models can be validated using existing RCT data, potentially slashing the cost and time of bringing AI diagnostic tools to market.
• AI-Designed Drugs in the Clinic: Rentosertib's advancement to Phase 2 provided the first clinical evidence that AI-designed molecules can work in humans.
• Human-in-the-Loop Frameworks: Research on robust human-AI collaboration provided practical guidance for deploying AI safely in clinical workflows.

Looking Ahead:

In 2026, we expect AI adoption to accelerate, particularly in data management, site operations, and clinical decision support. The winners will be organizations that focus not just on AI capabilities but on thoughtful integration into existing workflows with appropriate human oversight.

Related: Meet "Grace": How AI Agents are Revolutionizing Trial Recruitment

Regulators worldwide showed a remarkable willingness to adapt their frameworks to accommodate innovation in 2025. From digital health to AI to real-world data, agencies created new pathways designed to accelerate patient access while maintaining safety standards.

Key Developments:

• FDA TEMPO Program: Launched in December, this pilot program provides a faster pathway for digital health technologies in chronic disease, using enforcement discretion in exchange for real-world data collection commitments.
• EU COMPASS-AI: The EU's initiative provided practical guidance for deploying AI in clinical settings, with a focus on cancer care and underserved regions.
• Global AI/ML Guidance: Both the FDA and EMA issued additional guidance on AI/ML-based software as a medical device, clarifying expectations for continuous learning systems and algorithm change protocols.

Looking Ahead:

We expect continued regulatory evolution in 2026, with more pilot programs, guidance documents, and international harmonization efforts. Organizations should engage proactively with regulators and participate in pilot programs where appropriate.

Related: The TEMPO Pilot Program: A New Era for Digital Health in Chronic Disease

2025 saw AI-powered drug discovery move from theoretical promise to clinical reality. Multiple AI-designed molecules advanced in the clinic, and the ecosystem of AI drug discovery companies continued to grow and mature.

Key Developments:

• Clinical Milestones: Rentosertib's Phase 2a data showed efficacy signals in colorectal cancer, providing proof of concept for AI-designed small molecules. Other AI-originated compounds also advanced through clinical development.
• Partnerships and Acquisitions: Major pharmaceutical companies deepened their AI drug discovery partnerships, and several acquisitions signaled big pharma's commitment to the space.
• Expansion Beyond Small Molecules: AI began making inroads in biologics design, including antibody engineering and protein structure prediction, expanding the scope of AI-driven discovery.

Looking Ahead:

2026 will be a pivotal year for AI drug discovery. As more AI-designed drugs advance through trials, we will learn much more about whether the efficiency gains in discovery translate to improved clinical success rates. Early results are encouraging, but the ultimate test is clinical outcomes.

Related: The Rise of AI-Generated Drugs: Rentosertib and the Future of Precision Medicine

After years of talk about patient-centricity, 2025 brought tangible progress. Decentralized trial elements, digital endpoints, and patient engagement technologies moved from pilots to mainstream adoption.

Key Developments:

• Hybrid Trial Models: The fully decentralized trial remained rare, but hybrid models—combining traditional site visits with remote assessments, home health visits, and digital data collection—became the norm for many trials.
• Digital Endpoints: Regulators increasingly accepted digital endpoints derived from wearables and smartphones, particularly in neurological and musculoskeletal indications.
• AI for Engagement: AI-powered tools like Grace demonstrated that personalized, proactive patient engagement can significantly improve recruitment and retention.
• Patient Voice in Design: More sponsors engaged patient advocacy groups early in trial design, incorporating patient preferences into protocol development.

Looking Ahead:

In 2026, we expect patient-centric approaches to become table stakes. Trials that do not consider patient convenience and burden will struggle to compete for enrollment. The technology and operational infrastructure for decentralized elements will continue to mature.