Glatt Group

AI-PV Practical Series (From Human Oversight to HPVM) #1 What Became Clear Through the 1st HPC Study Group Meeting On 17-Apr-2026, we held the first HPC Open Study Group Meeting. In this small-group session, participants from pharma, technology providers, and consulting came together to discuss the current AI-PV landscape, CIOMS “Artificial Intelligence in Pharmacovigilance” (2025), and HPVM: HiroPharma Validation Method. What became clear to me once again is that the discussion around AI-PV is no longer simply about whether AI should be introduced. It is increasingly moving toward a more practical question: how to assure reliability, and how to establish accountability in real operations. In particular, there was strong interest in the concept of Human Oversight. AI is not intended to replace expert judgment, but to support it. If that is the case, then where should human involvement remain, what exactly should be reviewed, and how should AI-supported decisions be made explainable in practice? These appear to be central operational questions. At the beginning of the session, HPVM: HiroPharma Validation Method was still unfamiliar to some participants. By the end of the discussion, however, it seemed that HPVM was better understood as one possible practical approach for applying Human Oversight in AI-enabled pharmacovigilance. At the same time, many pharma and PV professionals are still at an earlier stage, asking a more basic question: how should we think about AI adoption in PV in the first place? That means there are still several steps of understanding between the principles articulated by CIOMS, the meaning of Human Oversight, and the need for a concrete implementation approach. In this New series, I would therefore like to explore the following six steps, one by one: Step-1: looking at the current reality of AI-PV Step-2: identifying what is really at stake in AI adoption Step-3: clarifying what CIOMS AI in PV 2025 has actually shown Step-4: recognizing why Human Oversight cannot remain only a concept Step-5: identifying what MAHs must design for themselves in practice Step-6: considering why an implementation-oriented approach such as HPVM becomes necessary The discussion on AI-PV is moving from adoption itself to the design of operation, oversight, and accountability. I believe we are now entering that stage. #Pharmacovigilance #AI #AIPV #HumanOversight #HPVM #HiroPharmaValidationMethod #CIOMS #DrugSafety #PV #AIValidation #RegulatoryScience

Yesterday, I attended BREAKTHROUGH 2026 APAC Tokyo. It was very valuable to hear practical perspectives from companies and stakeholders already working with AI-enabled PV operations. What particularly stood out to me was that several speakers referred to Human Oversight and HITL in the context of CIOMS “Artificial Intelligence in Pharmacovigilance” (2025). This gave me the impression that CIOMS AI in PV 2025 is beginning to be recognized as an important practical reference document in current AI-PV discussions. One practical point that stayed strongly with me was the design scope of [Human Oversight] in Case Intake & Triage. In some of the operational examples shared at the meeting, cases identified by AI as important were subject to [HITL: Human-in-the-Loop] QC review. This seems to be one realistic approach to combining AI and human review in order to support reliability. At the same time, I felt that the handling of cases assessed by AI as lower priority remains an important open question. In pharmacovigilance, rare events and outliers can be intrinsically important, so the real issue may not be only whether [Human Oversight] exists, but also how broadly and how systematically it is designed. At present, dedicated Japan-origin AI-PV guidance is still limited. For that reason, I believe the importance of internationally structured reference documents such as CIOMS AI in PV 2025 will continue to grow. This event reminded me that AI-PV has already entered a real operational phase, and that CIOMS AI in PV 2025 is beginning to serve as one of the foundations for ongoing practical discussion. #BTAPAC26 #Pharmacovigilance #AI #AIPV #HumanOversight #CIOMS #DrugSafety #PV #AIValidation #RegulatoryScience #LifeSciences