超越自动化：人工智能时代战略性LQPPV外包如何保障PV合规性

Beyond Automation:How Strategic LQPPV OutsourcingFuture-proofs PV Compliancein the AI Era ANA PEDRO JESUÍNO, Associate Marketed Product Safety DirectorDOMINIQUE COLEMAN, Senior Director, Operations Lifecycle SafetyBARRY MULCHRONE, Head of Safety Science & Strategy, Lifecycle Safety Table of contents Introduction1What regulatory changes are impacting the PV climate?2Where does AI fit into PV workflows?2Why is outsourcing LQPPV work worthwhile?3How do I enhance my PV offerings?4About IQVIA4About the authors5 Introduction The field of pharmacovigilance (PV) continues to rapidly evolve due to changingregulations, increased adverse event reporting across multiple platforms,and the global expansion of pharmaceutical development. In response,Marketing Authorization Holders (MAHs) are exploring how to integrateartificial intelligence (AI) into their workflows to increase efficiency and lessenthe burden on human experts. Yet, despite growing AI usage across PV,the role of Local Qualified Persons for PV (LQPPVs) — who serve as the MAHrepresentatives and primary contacts for local health authorities in theirdesignated country — remains essential and pivotal. Amid this delicate landscape, the question of whetherto outsource LQPPV responsibilities is top of mind.Recent European Medicine Agency (EMA) guidanceupdates have emphasized the importance of robustvendor oversight and, in response, MAHs areconsidering which partnerships will best facilitatestreamlined vendor management. Strategically selectingan outsourcing partner with a wide array of PV services,advanced technology offerings, and ample regulatoryknowledge will allow you to leverage their expertiseto ease operational burdens, align with industry bestpractices, and maintain consistent compliance withevolving PV and AI guidelines. Strategically selecting an outsourcing partner with a widearray of PV services, advancedtechnology offerings, and ampleregulatory knowledge will allowyou to leverage their expertiseto ease operational burdens. What regulatory changes areimpacting the PV climate? Where does AI fit intoPV workflows? Throughout PV workflows, the processing and analysisof vast quantities of data stand as pivotal tasks thattraditionally burden human PV experts extensively.Traditionally, this data is manually analyzed by humanPV experts, which can be a major strain on time andresources. However, AI advancements like machinelearning and automation are being implemented toenhance the efficiency and accuracy of data processingfor PV. AI technology can analyze adverse event data andrapidly identify, ingest, and report safety signals, ensuringtimely responses that trigger necessary safety actions. Both the EMA good pharmacovigilance practices(GVP)and CommissionImplementing Regulation (EU)520/2012 guidelines have undergone recent updatesthat address the importance of an MAH’s commitmentto strengthening their protocols for vendor oversightand management. MAHs typically employ a broadarray of vendors across functions, including aggregatereporting, risk mitigation, social media monitoring,market research, and various other functions, whichadds complexity to the vendor oversight process.Both documents implore MAHs to establish rigorousvendor management protocols that ensure a vendoris compliant, their workload is under control, and anMAH is apprised of any issues that arise throughouta partnership. AI advancements like machinelearning and automation arebeing implemented to enhancethe efficiency and accuracyof data processing for PV. Satisfying these goals requires MAHs to haverobust procedures in place that provide clear anddetailed expectations when it comes to a vendor’sresponsibilities, scope of work, and escalation path.Each of these components should be thoroughlydetailed and documented so that both MAH and vendorcan guarantee compliance when facing inspection. Though adding AI into previously human-only processesmay provoke skepticism from stakeholders, the goal is tobetter support human experts rather than replace them,creating a synergistic approach that maximizes thebenefits of both humans and AI. AI implementation in PVaims to reduce some of the manual processing burden,enabling experts to focus their knowledge where it ismost beneficial. There are also certain components ofPV processes that cannot be replaced by AI. For example,AI cannot adequately detect human emotions. When itcomes to adverse event intake, identifying the emotion In addition to thorough expectations around vendormanagement, regulatory authorities are also consideringhow best to integrate and regulate AI in processesacross the drug development lifecycle. The EMA hasmade several updates to support its integration,including areflection paper on AIthat discusses AIand machine learning applications across the medicinalproduct lifecycle, an update on AI in PV that highlightsAI’s role in enhancing adverse event report managementand signal detection, and the GVP, which providedupdat