Artificial intelligence (AI) is revolutionising industries worldwide, including pharmaceuticals, where it’s driving significant advancements in drug development and clinical trials. Real-world examples illustrate these advancements. Insilico recently announced the successful progression of its lead candidate to Phase IIb trials, while Isomorphic Labs and Recursion have achieved significant milestones in AI-enabled drug development. These successes demonstrate how AI is reshaping the drug discovery landscape, accelerating research and introducing new drug candidates into the research pipeline.

AI in Clinical Trials: From Hype to Reality

While much of the conversation around AI focuses on its potential, tangible benefits are also being realised in clinical trials. By automating complex processes and enhancing data-driven decision-making, AI is shifting from theoretical promise to practical application. AI is streamlining workflows, accelerating timelines, and enhancing operational consistency in ways that were previously unattainable. Practical examples include automation of repetitive tasks such as protocol development, database build, data cleaning, and report generation. This transformation is being led by technology-enabled contract research organisations (CROs), which are harnessing AI to improve speed, efficiency, quality, and ultimately patient outcomes in clinical trials.

Streamlining Trial Design and Execution with AI

One of AI’s most impactful applications is protocol development and optimisation. Large language models (LLMs) are particularly adept at automating document-intensive tasks, such as creating initial drafts of protocols and informed consent forms. By analysing historical data and using standardised templates, LLMs can generate high-quality drafts that human editors refine, allowing teams to focus on trial design.

AI-driven protocol digitisation takes this further by transforming unstructured protocols into structured formats that are machine- and human-readable. This enables the automation of critical database build elements, such as case report forms (CRFs), and edit checks. The result is a reduction in electronic data capture (EDC) build times by up to 30%, allowing databases to be ready as soon as sites are activated. And by standardising data according to frameworks like Clinical Data Acquisition Standards Harmonisation (CDASH), these tools eliminate the need for time-consuming data mapping, ensuring trials start on schedule.

Optimizing Site Selection and Patient Recruitment

Once the trial design is finalized, site selection and patient recruitment become critical. These stages are often bottlenecks in the clinical trial process, as inadequate site performance or poor recruitment rates can lead to delays or even trial failure.

AI-powered tools are transforming site feasibility assessments by analysing historical site performance data and aligning it with protocol requirements. This enables sponsors and CROs to identify optimal sites for their trials. However, current AI tools face limitations due to insufficient access to aggregated data, highlighting the need for robust, privacy-protected data pipelines. By integrating publicly available datasets and proprietary information, AI can further improve site selection accuracy and enrolment predictions.

On the recruitment side, AI can analyse patient eligibility criteria and match them with electronic health records (EHRs), streamlining the recruitment process. This not only accelerates enrolment but also ensures the right participants are selected, improving trial outcomes.

Enhancing Risk-Based Quality Management

AI is also proving indispensable in risk-based quality management (RBQM), where it identifies potential risks and anomalies in real time. During the H1N1 pandemic in 2009, Emmes used early versions of risk-based monitoring technology. Today, AI has elevated these capabilities, enabling dynamic adjustments to monitoring strategies based on site and patient data. This ensures that resources are allocated effectively, prioritising high-risk areas and enhancing trial integrity.

Fraud detection is another area where AI excels. By analysing enrolment patterns and cross-referencing data across sites, AI can identify “professional patients” or subtle inconsistencies that may compromise trial results. These tools enhance data quality, leading to more reliable outcomes and higher standards of study integrity.

Streamlining Data Analysis and Study Closeout

As trials conclude, AI accelerates data reconciliation, analysis, and reporting, ensuring that study results are delivered to regulators efficiently and accurately. A key application is the automation of data standardization into relational formats using CDASH and SDTM-controlled terminology. These capabilities allow up to 95% of the mapping required for FDA submissions to be automated, drastically reducing database lock timelines.

In addition to standardising data, AI is revolutionising the generation of Tables, Figures, and Listings (TFLs) and Clinical Study Reports (CSRs). By leveraging structured data and predefined templates, AI can automate the creation of these critical regulatory documents. This automation reduces the time required to compile, format, and validate TFLs and CSRs, ensuring compliance with regulatory standards while accelerating submission timelines. Interactive reporting pipelines further enhance this process, enabling the rapid production of high-quality outputs and facilitating earlier decision-making.

The reusability of standardised data and code for TFLs and CSRs ensures that lessons from previous trials are applied consistently, improving both efficiency and quality. By cutting the time to generate and review these documents, sponsors and CROs can meet regulatory deadlines more easily and bring life-saving treatments to market faster.

Unlocking the Potential of Agentic AI

As AI evolves, agentic AI—AI systems capable of taking autonomous, goal-directed actions—holds the potential to redefine clinical trial execution. Unlike current AI tools, which require predefined instructions or prompts, agentic AI systems will dynamically prioritize tasks, learn from outcomes, and adjust in real time.

For clinical trials, agentic AI could enhance execution in several critical ways:

• Adaptive Protocol Execution: Agentic AI systems could continuously monitor trial progress and adjust protocols dynamically based on emerging data, such as amending inclusion/exclusion criteria or modifying visit schedules to improve retention.
• Proactive Risk Management: By autonomously detecting anomalies, agentic AI could implement preemptive mitigation strategies without human intervention, minimising disruptions and ensuring data quality.
• Resource Optimisation: These systems could analyze workloads and staff availability to allocate resources in the most effective manner, improving efficiency and reducing costs.
• Participant Engagement: By interacting directly with participants via intelligent chatbots or virtual assistants, agentic AI could provide personalised support, answer questions, and send reminders, improving retention and compliance.

The potential of agentic AI lies not only in its ability to automate processes but also in its capacity to make intelligent, context-aware decisions. As this technology develops, its integration into clinical trials will enable more adaptability, efficiency, and patient-centricity, ultimately driving better outcomes.

Preparing for the Future of AI in Clinical Trials

The integration of AI into clinical trials is still in its early stages, but its potential is immense. Over the next five years, AI is expected to:

• Design better studies with reduced participant and site burdens.
• Enhance real-time monitoring and risk management.
• Automate the drafting of key documents, such as protocols, patient profiles and clinical study reports.

To fully realise these benefits, the industry must address key challenges, including the need for robust data pipelines, ethical AI design, and regulatory compliance. Incorporating human oversight and engineered prompts ensures that AI tools remain accountable and trustworthy, aligning technological innovation with patient safety and trial integrity.

By embracing AI, CROs and sponsors can not only accelerate clinical trials but also enhance their quality, ultimately bringing life-saving treatments to patients faster and more effectively.

Noble Shore serves as the Vice President of Technology Strategy & Product Adoption at Veridix AI, a subsidiary of the Emmes Group. With over 20 years of experience in the pharmaceutical and biotechnology industries, Noble has led cross-functional teams to drive innovation in patient recruitment, data management, and operational efficiencies. Before joining Veridix AI in January 2024, Noble held various leadership positions at Emmes, including Vice President of Software Development, where he designed comprehensive product strategies, and Associate Vice President of Software Development, focusing on cloud transformation of the Advantage eClinical software. His tenure at Emmes began in November 2005, during which he also served as Director of Software Development, Software Engineering Manager, and Senior Programmer/Analyst, contributing to major software launches and engineering transformations.

Henry Lin is currently Senior Vice President, Data Science & AI at Emmes Group. Henry and his team build scalable data science and engineering software and services that are integrated into our platform to accelerate the generation of actionable insights across the clinical development lifecycle. Previously, Henry has led data science teams across different functions at Medidata Solutions, Janssen Pharmaceuticals, Johnson & Johnsons Medical Device Companies and Roche Pharmaceuticals. He earned his B.S. in Electrical Engineering & Computer Science from UC Berkeley and Ph.D. in Biological & Medical Informatics from UCSF.