With applications abound, AI is poised to change how RWD/RWE teams operate and how quickly they can accelerate drug development. Despite the excitement over various use cases for AI, regulatory requirements and risk mitigation strategies leave many pharmaceutical and biopharmaceutical teams questioning where to start, looking for clear regulatory guidance, but very interested in how to shorten their timelines. The blog below addresses some of these concerns and gives epidemiology teams a place to start.
AI has been applied to various steps in the drug development process and holds significant untapped potential through integration with RWD/RWE. Applications of AI with RWD include optimization of clinical trial cohorts, analysis of complex medical data, extraction of insights from unstructured clinical data (such as notes from electronic health records), and improvement of the quality and validity of research.
Despite the excitement around AI, there are still concerns about bias and the interpretation of results. This will continue to be a significant hurdle in the future, and organizations must address these issues before fully integrating AI into their workflows.
Another hurdle with implementing AI tools, like large language models (LLMs,) is ensuring the privacy of personal health data. Using synthetic data and automating de-identification with AI can help with this task, enabling compliance with privacy laws, such as GDPR or HIPAA, and preventing sensitive data from being remembered or used for training by LLMs.
RWD/RWE teams should create transparent and standardized protocols for how AI tools are used with certain data types and how patient privacy is protected.
Generative AI provides a lot of potential use cases for RWE and the broader pharmaceutical landscape.
To zero in on where to start, choose a small, manageable, and repetitive task where you can use an LLM-based chatbot to help you draft a document. From there, you can expand to higher-impact use cases over time. AI can streamline routine tasks like quality assurance, medical writing, and data management. Chatbots and virtual assistants can also enhance data capture during patient visits, while automation can ease the process of converting RWD into formats required for regulatory submissions.
Identifying repeatable, high-value tasks ensures consistent returns and builds confidence across RWD/RWE teams and organizations.
The evolving regulatory landscape creates uncertainty about AI’s role in drug development and how regulatory bodies will react to its use.
While this may cause some risk-averse organizations to steer clear of using AI, now is the right time to bring AI use cases to the FDA since there is a lot of public excitement and attention in the area. There is likely a draft guidance in the works for AI/ML in drug development, but in the meantime, there is a discussion paper on this topic and AI use in drug manufacturing. If your organization has a unique AI/ML use case, you should engage early with regulators to ensure compliance with the current regulatory uncertainty.
Despite its benefits, AI introduces risks, such as hallucinations and generating unreliable outputs. Organizations need processes to assess AI tools and ensure safe experimentation. To do this, rely on the experience of senior personnel for careful and comprehensive vetting of new AI tools. Experienced professionals should also encourage and mentor younger colleagues, fostering the skills to navigate AI-generated insights with accuracy and compliance in mind.
AI presents enormous potential for RWD/RWE teams to transform their operations, but thoughtful planning, regulatory engagement, and risk management are essential to success. By starting small and expanding gradually, organizations can embrace AI to enhance efficiency, maintain compliance, and drive innovation in drug development.
To learn more about Aetion and how we’re integrating the latest computational tools into our software platform, contact us today.