Applied AI and Advanced Automation in Clinical Trials

Kirk Wroblewski, Chief Information Officer, ProPharma

Pharmaceutical services companies and now contract research organizations (CROs) currently find themselves as key members of the knowledge community. The nature of the work is not only information-based but also consistently nonlinear, only partially predictable, networked, interdependent and increasingly collaborative.

The life sciences industry is on the cusp of a technological revolution, driven by the rapid advancements in artificial intelligence (AI). While much attention has been given to the role of AI in drug discovery, an equally transformative application is unfolding in the auxiliary areas of life sciences.

This article will discuss:

• How advances in AI and generative artificial intelligence (GenAI) solutions provide a key to unlock productivity and fundamentally help the industry move from volume to value.
• Concrete examples of current challenges being addressed by GenAI in the clinical development space.
• Observations on practices to implement for more success in the GenAI space.
• Future innovations in GenAI.

The integration of GenAI into these areas is not just a technological upgrade; it represents a paradigm shift in how life sciences operate. For professionals in the field, understanding and leveraging GenAI can lead to better decision-making, streamlined processes, and ultimately, improved patient outcomes. With the industry grappling with ever-increasing data volumes and complexity, GenAI offers a solution that goes beyond traditional analytics, enabling more nuanced and sophisticated analysis.

In order to better understand GenAI opportunities in the CRO life science space, let’s dig into some emerging use cases in medical information (MI), pharmacovigilance, and clinical studies/CRO applications.

Medical information: Automating insight generation

MI within CROs plays a crucial role in the life sciences and drug development process, acting as a bridge between scientific data and healthcare professionals (HCPs) or patients. These HCPs are responsible for managing inquiries related to drug safety, efficacy, and clinical trial data. They ensure that all medical information provided is accurate, compliant with regulatory guidelines, and supports the safe use of investigational drugs. This practice is essential in creating transparency and trust between the pharmaceutical company, medical community, and patients.

GenAI is making significant inroads in the work of MI teams within CROs. Currently, GenAI can assist in tasks such as automating responses to frequently asked medical queries, generating summaries of clinical trial results, and processing large volumes of regulatory documents. By leveraging vast datasets, GenAI enhances the speed and accuracy of literature reviews, ensuring that information provided to HCPs and patients is both current and evidence-based. These capabilities help MI teams focus on more complex inquiries while improving overall operational efficiency.

In the future, GenAI’s role in MI is expected to expand, potentially taking on more nuanced tasks such as personalizing medical information for specific patient groups or clinical settings. GenAI models could analyze adverse event reports and provide early warnings about emerging safety trends or regulatory risks. Additionally, GenAI’s ability to continuously learn from new data will enable CROs to maintain the most up-to-date knowledge, helping them stay ahead of evolving regulatory landscapes and enhancing their support for clinical trials and drug safety monitoring.

Pharmacovigilance: Enhancing drug safety

Pharmacovigilance is the process of monitoring the safety of pharmaceutical products once they hit the market. The challenge here is the sheer volume of adverse event reports that need to be analyzed to detect potential safety signals.

Across MI and pharmacovigilance, GenAI automates the analysis of vast data sets, delivering more accurate and timely insights while enhancing patient safety. GenAI is stepping up to this challenge by automating the detection and analysis of these signals. AI models can rapidly process and analyze adverse event data, identifying patterns that might indicate a safety issue. This accelerates response times and improves accuracy, ensuring that medications on the market remain safe for patients. AI algorithms that monitor social media, forums, and other digital platforms for mentions of drug side effects, can provide early warning of potential safety issues. GenAI is also enhancing the ability to detect adverse drug reactions by automating the analysis of large pharmacovigilance datasets. This leads to faster and more accurate identification of safety signals, ultimately improving drug safety.¹

Drug regulatory submissions: Accelerating approvals

Submitting regulatory documents for new drugs is often a bottleneck in bringing therapies to market. The preparation of these documents is highly complex, requiring meticulous attention to detail and strict adherence to regulatory guidelines.

GenAI is revolutionizing this process by automating the generation of regulatory documents. AI models can ensure that documents are correctly formatted, comprehensive, and compliant with regulatory standards, reducing the risk of delays due to errors or omissions. This speeds up the approval process, enabling new therapies to reach patients faster.²

CRO operations: Streamlining clinical trials

CROs are essential partners in the clinical trial process, handling everything from patient recruitment to data management. However, the increasing complexity of clinical trials presents significant challenges, particularly in terms of managing and integrating large volumes of data from multiple sources.

GenAI is providing solutions by automating key aspects of clinical trial management. For example, AI models can optimize patient recruitment by analyzing data to identify the most suitable candidates for trials. They can also ensure that data from different trial sites is consistent and integrated, improving the reliability of trial results.

GenAI systems and tools that predict patient dropout rates in clinical trials allow CROs to take proactive measures to retain participants and ensure trial success.

Avoiding the pitfalls in using GenAI

The growing hype around GenAI has raised concerns, particularly about the risk of misapplying the technology without careful consideration of the right use cases or operational needs. GenAI, while powerful, is not a one-size-fits-all solution, and implementing it without a clear strategy can lead to inefficiencies, ethical concerns, or even operational failures. Many life sciences organizations may rush to adopt GenAI without fully understanding its limitations, such as bias in model outputs, lack of transparency, or data privacy issues. When not aligned with clear business objectives or properly integrated into workflows, GenAI can create more problems than it solves, potentially leading to inaccurate results or regulatory risks.

Start with human-augmented GenAI use cases to minimize risk

Human-augmented GenAI is an approach where AI models work alongside human professionals to generate outputs that require human validation before implementation. This human-in-the-loop framework ensures that the AI-generated content or decisions are accurate, contextually appropriate, and aligned with ethical standards. By incorporating human oversight, this method mitigates risks associated with fully autonomous AI systems, which may otherwise produce biased, incomplete, or erroneous results. This is especially crucial in regulated industries such as clinical research or financial services, where even small errors could lead to major compliance issues or endanger lives.

Future Innovations in GenAI

Leverage fine-tuned large language models to improve GenAI success

One key challenge in implementing AI in life sciences is the need for models tailored to specific datasets. Generic models often lack the specificity required for accurate analysis in specialized areas such as MI or pharmacovigilance. The future will likely see the development of fine-tuned GenAI models that are trained on specific data sets, improving their accuracy and relevance.

Evolution of RAG architecture: Contextual AI

Retrieval-augmented generation (RAG) architecture represents a cutting-edge approach that combines the strengths of retrieval-based and generative AI models. As RAG technology evolves, it will enable AI systems to provide more contextually accurate and relevant outputs, particularly in areas where context is critical.

In the realm of drug regulatory submissions, for example, RAG models could be used to generate documents that not only comply with regulatory standards but also incorporate the most relevant and up-to-date scientific information. This would ensure that submissions are not only accurate but also highly relevant, reducing the risk of regulatory pushback.

Graph databases and GNNs: Enhancing contextual understanding

The use of graph databases and graph neural networks (GNNs) represents a cutting-edge approach to injecting context into AI models. By mapping relationships between different data points, these technologies can enhance the contextual understanding of complex datasets, leading to more accurate and insightful analyses. This will be particularly valuable in areas such as pharmacovigilance and MI, where understanding the broader context is often key to making informed decisions.³

Understanding when and how to engage

Engaging and leveraging GenAI requires a strategic approach to effectively solve challenges within the life sciences sector. Organizations should begin by identifying specific pain points in their workflows, such as inefficiencies in data processing or gaps in knowledge extraction. Once these areas are pinpointed, they can seek out GenAI partners or internal teams equipped with the necessary expertise to tailor solutions to their unique needs. It’s crucial to involve stakeholders early in the process to ensure alignment on objectives and outcomes. As projects progress, continuous feedback loops can help refine the models and ensure they are delivering the expected value.

Embracing the GenAI revolution

The integration of GenAI into auxiliary areas of life sciences is not just a technological advancement but a necessary evolution to keep pace with the growing complexity and scale of data in the industry. As GenAI continues to evolve, its role in transforming clinical operations will only grow more essential, driving efficiency, improving regulatory compliance, and ultimately enhancing patient care across the industry.

Kirk Wroblewski is Chief Information Officer, ProPharma

References

1. Praveen, J.; Kumar Cm, K.; Haralur Channappa, A. Transforming Pharmacovigilance Using Gen AI: Innovations in Aggregate Reporting, Signal Detection, and Safety Surveillance. J. Multidiscip. Res. 2023. 3 (3), 9-16. https://www.researchgate.net/publication/374535382_Transforming_Pharmacovigilance_Using_Gen_AI_Innovations_in_Aggregate_Reporting_Signal_Detection_and_Safety_Surveillance

2. Niazi, S.K. The Coming of Age of AI/ML in Drug Discovery, Development, Clinical Testing, and Manufacturing: The FDA Perspectives. Drug Des Devel Ther. 2023. 9 (17), 2691-2725. https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10493153/

3. Alcaraz, A. LLM and GNN: How to Improve Reasoning of Both AI Systems on Graph Data. Towards Data Science. December 3, 2023. https://towardsdatascience.com/llm-and-gnn-how-to-improve-reasoning-of-both-ai-systems-on-graph-data-5ebd875eef30