| Q1 : What is HPVM? | |
| A1 | HPVM（HiroPharma Validation Method）は、医薬品安HPVM (HiroPharma Validation Method) is a methodological framework designed to ensure the reliability and transparency of AI functions used in Pharmacovigilance (PV) activities. It was developed by HiroPharmaConsulting Co., Ltd. and is protected by Japanese Patent No. 7778327. |
| Q2 : Does HPVM depend on any specific AI model or vendor? | |
| A2 | No. HPVM is a vendor-agnostic and model-agnostic conceptual framework. It can be applied to any AI model regardless of algorithm, architecture, or supplier. |
| Q3 : What does HPVM define? | |
| A3 | HPVM defines an evaluation framework, not prescriptive requirements. Specific thresholds, acceptance criteria, evaluation frequency, and operational rules remain the responsibility of each Marketing Authorization Holder (MAH) based on risk-based judgment. |
| Q4 : Does HPVM guarantee AI performance? | |
| A4 | No. HPVM does not guarantee AI performance. Instead, it provides a structured approach to monitor performance variability, ensure transparency, and support explainability over time. |
| Q5 : How does HPVM align with GVP? | |
| A5 | HPVM supports the core intent of GVP, including accuracy, reliability, reproducibility, and transparency, by providing a practical framework for managing AI behavior in routine PV operations. |
| Q6 : Can HPVM be applied to black-box AI models? | |
| A6 | Yes. HPVM evaluates outputs, stability, differences, and traceability, rather than internal model logic. Therefore, it can be applied to black-box AI systems. |
| Q7 : What is the √N+1 sampling approach? | |
| A7 | √N+1 is one example of a commonly used statistical approach to ensure representativeness of a population. It is not a regulatory requirement, but a rational method suited to evaluating AI performance variability. |
| Q8 : Does HPVM require regulatory approval or endorsement? | |
| A8 | No. HPVM does not seek regulatory approval, endorsement, or recommendation. It is a supplementary framework intended to support responsible implementation under existing regulatory principles. |
| Q9 : What does Japanese Patent No. 7778327 protect? | |
| A9 | The patent protects a method for AI reliability assurance in Pharmacovigilance, in which use (PV), regulatory context (GVP/CSV), and evaluation structure are combined into an integrated validation methodology. |
| Q10 : Can HPVM concepts be applied outside the pharmaceutical industry? | |
| A10 | Yes. The general concept of AI reliability assurance may be applied in other industries such as finance or legal services. Such general applications are not the direct target of this patent, which is focused on the PV domain. |
| Q11 : When could patent infringement occur? | |
| A11 | Infringement may occur if: ・The method is implemented within a Pharmacovigilance context, and ・The same conceptual approach and structural elements defined in the patent claims are substantially implemented, and ・The method is used commercially without authorization. Changing the name or labeling of the method does not eliminate infringement if the substance of the method is the same. |
| Q12 : Is a license required? | |
| A12 | For pharmaceutical companies (MAHs), a conditional no-cost license is planned to prioritize public health and patient safety. For PV system vendors and service providers, separate licensing arrangements may apply. |
| HPVM (HiroPharma Validation Method) is protected by Japanese Patent No. 7778327. | |
| Copyright© 2025 HiroPharmaConsulting® Co., Ltd. All rights reserved. HPVM (HiroPharma Validation Method) is protected by Japanese Patent: Patent Number: Japan Patent No. 7778327/ Grant date: 21 November 2025. This document contains proprietary information. Unauthorized reproduction or disclosure is strictly prohibited. |
PV-AI HPVM : PV-AI HiroPharma Validation Method — FAQs — (English) 英語
