Enhancing Drug Safety with AI and Automation Technologies - with Marie Flanagan of IQVIA
🎯 Summary
Podcast Episode Summary: Enhancing Drug Safety with AI and Automation Technologies - with Marie Flanagan of IQVIA
This 22-minute episode of the AI and Business Podcast, featuring Marie Flanagan, Director of Product Management at IQVIA, focuses on the critical role of AI and automation in transforming drug safety and pharmacovigilance. The discussion highlights how established AI disciplines, particularly Natural Language Processing (NLP), are being strategically combined with automation tools to improve adverse event (AE) detection, especially from unstructured and non-traditional data sources like social media.
1. Focus Area
The primary focus is the application of AI and automation technologies—specifically NLP, Robotic Process Automation (RPA), and Optical Character Recognition (OCR)—to enhance adverse event detection and safety signal monitoring within the pharmaceutical industry. The conversation emphasizes moving from reactive to proactive safety workflows.
2. Key Technical Insights
- Strategic Combination for Efficacy: The most favorable outcomes in adverse event detection (e.g., achieving 70-80% positive outcomes from social media data) are achieved not by using pure AI alone, but by mixing NLP/AI with automation tools like RPA and OCR to process and structure complex, unstructured data.
- NLP as the Foundation: IQVIA has utilized proprietary, safety-trained NLP for over a decade (in their Vigilance Detect product) to find safety signals upstream of traditional workflows, noting that modern Large Language Models (LLMs) fall under the broader NLP discipline.
- Automation as the “Hands”: RPA is described metaphorically as the “hands” that execute the commands derived from the “brain” (AI/NLP), emphasizing that while RPA is not strictly AI, it is essential “engine oil” for operationalizing AI insights in complex safety processes.
3. Business/Investment Angle
- Social Media Maturation: Social media, once dismissed as a source of meaningful safety data, is now becoming valuable for pre-signaling—detecting emerging health trends (like localized flu outbreaks) months before they appear in formal regulatory systems (like FAERS).
- Regulatory Collaboration: The dynamic between regulators (FDA, EMA) and industry is shifting toward collaboration. Regulators are encouraging digitalization and are primarily focused on verifying that companies have robust internal checking mechanisms (“They’re checking that you’re checking”).
- Workflow Optimization is Key: Successful AI implementation hinges on first analyzing and optimizing existing workflows. Projects fail when the underlying process is not fit for digital enhancement.
4. Notable Companies/People
- Marie Flanagan (IQVIA): Guest and expert, providing insights from her role in product management for digital solutions, particularly IQVIA’s Vigilance Detect tool.
- IQVIA: The company whose long-standing use of NLP in pharmacovigilance serves as the central case study.
- Regulators (FDA, EMA): Mentioned as actively encouraging digital enhancement and setting broad, non-prescriptive guidelines for AI implementation.
- C-M-O-S-I-X Working Group: Mentioned as an influential industry body laying out specifics for the safety sector.
5. Future Implications
The industry is moving toward a patient-centric, proactive pharmacovigilance model. By leveraging multi-channel data collection (including social media), automation frees up human agents to focus on empathy and gathering more clinically robust information, potentially allowing the industry to prevent adverse events rather than just react to them.
6. Target Audience
This episode is highly valuable for Life Sciences and Healthcare Executives, Product Managers, Regulatory Affairs Professionals, and Technology Leaders involved in AI investment, strategy, and deployment within pharmaceutical and biotech companies. It is relevant for those navigating real-world AI adoption challenges.
Comprehensive Narrative Summary
The podcast episode provided a deep dive into how IQVIA leverages a sophisticated blend of AI and automation to modernize drug safety monitoring, moving beyond traditional, reactive workflows. Marie Flanagan established that NLP has been a core component of IQVIA’s safety detection tools for over a decade, sifting through vast, multilingual, unstructured datasets to identify adverse events upstream.
A central theme was the synergistic power of combining different technologies. Flanagan stressed that while NLP is the “brain,” automation tools like RPA (the “hands”) and OCR are necessary to achieve high-fidelity results when processing complex data from sources like social media. Without this integration, NLP alone yields lower success rates (20-30%), whereas the combined approach boosts positive outcomes significantly (up to 70-80%).
The discussion then pivoted to the regulatory landscape. Flanagan noted a positive shift where regulators are encouraging digitalization and adopting a more collaborative stance. The emphasis is now on industry self-governance; regulators are focused on auditing the processes companies use to ensure safety, rather than dictating granular technical specifications. This collaborative environment supports the move toward Real-World Evidence (RWE) gathering.
Furthermore, the integration of technology is changing the role of the “human in the loop.” By automating tedious data collation, agents gain time to exercise empathy and gather richer, more clinically robust information directly from patients via new digital channels. This enables the exciting future possibility of proactive pharmacovigilance—predicting and preventing safety issues before they escalate. Flanagan concluded with actionable advice for leaders: “Start small, but start,” prioritize workflow assessment before implementation, and seek high-value, tactical changes rather than waiting for perfect, large-scale AI solutions.
🏢 Companies Mentioned
đź’¬ Key Insights
"Here's another. The regulators aren't checking you. They're checking that you're checking. That's where the cooperative environment is heading."
"The regulators aren't checking you. They're checking that you're checking. That's where the cooperative environment is heading."
"Our best projects with the most favorable outcomes come from mixing RPA, OCR, traditional machine learning, and advanced AI. That's where the magic happens."
"look at your workflow first. Look at the most practical steps you can take to implement digital enhancement in your workflow. Unfortunately, we've seen projects fail and AI projects fail because the workflow has not been conducive to it in successful."
"We can now take that information and use it for some pre-signaling. So we've seen some really great examples how you can hotspot various incidences of influenza arising in certain suburbs in Manhattan, for example, and then correlate that data to what might exist in the FDA's FAERS system a couple of months later."
"we could prevent adverse events happening rather than react to adverse events happening. Something we talk a lot about on that show, that difference between reactive and proactive workflows."