Clinical Trial Data Quality Hinders AI Adoption

What happened

The clinical research industry faces significant hurdles in adopting artificial intelligence (AI) due to issues with data quality and accessibility. While AI offers the potential to identify patterns, optimize study designs, and accelerate drug development, its effectiveness is directly tied to the quality and timeliness of the data it receives.

Current data practices are often fragmented and involve lengthy reconciliation processes, which are not compatible with AI systems that require continuous data streams for real-time decision-making. Experts suggest a shift towards prioritizing data quality at the source and ensuring real-time data access, a concept referred to as becoming "data-native" before "AI-native."

Concerns also exist regarding the sheer volume of data collected in clinical trials, with research indicating that a substantial portion, nearly 30%, does not directly inform key decisions. Since 2005, the number of procedures per protocol has increased by nearly 140%, endpoints by over 200%, and data points collected by more than 600%, suggesting that an "more is better" approach to data collection may be counterproductive.

Why it matters

High-quality, well-governed, and real-time data is emerging as a critical strategic asset. It enables better decision-making, supports automation, reduces operational inefficiencies, and enhances the effectiveness of AI-driven insights. Without addressing data quality and accessibility, the full benefits of AI in accelerating drug development and improving trial efficiency cannot be realized.

Key context

AI can identify patterns in large datasets, optimize study designs, predict enrollment challenges, detect safety signals earlier, and potentially speed up the development of new therapies. However, AI systems' output quality deteriorates rapidly if the data is incomplete, delayed, or inaccurate. The FDA has published guidance on Real-Time Clinical Trials (RTCT) to promote modernized infrastructure and more timely data access for real-time decision-making. Organizations that successfully leverage AI often focus on preventing errors rather than correcting them, aligning with quality-by-design principles and emerging regulatory expectations like ICH E6(R3).

What to watch next

Questions remain about how the industry will transition to a "data-native" approach. It is unclear how organizations will implement strategies to ensure data quality at the source and achieve real-time data access. Additionally, the specific methods for reducing the volume of collected data while ensuring it directly informs key decisions are yet to be fully defined.