In Japan, the basic principles of using artificial intelligence for participants in the pharmaceutical industry were presented: the document defines approaches to the use of neural networks in the development and evaluation of drugs The Center for Regulatory Science of the Japanese Medicines and Medical Devices Agency (PMDA) published a guide, "Guidelines for the use of artificial intelligence in regulatory activities (preliminary consideration)", which sets out approaches to the responsible use of artificial intelligence (AI) in the pharmaceutical industry. The document covers the areas of drug development, evaluation and regulation, including the use of neural networks to support regulatory judgments. It is mainly aimed at pharmaceutical companies, sponsors of clinical trials, developers and suppliers of AI models, as well as the regulators themselves. The guide describes the possibilities of using neural networks, including generative AI, during development, preclinical and clinical studies, registration and manufacturing. The document does not impose strict requirements on participants in the pharmaceutical market - it only specifies general principles for the responsible use and development of AI. PMDA proposed a three-tier structure. The human-centered principle comes first, followed by six basic principles and five areas of application and management. * So, AI cannot replace human judgments and decisions, but only support them. It is not acceptable to rely only on the neural network - the main responsibility will remain with the person. The six core principles include: A risk-based approach - the control measures must be appropriate to the level of risk. Correct definition of the context and the scope of the application - the goals of using AI, the accuracy of the results and the limits of the application must be determined in advance. Data management - it is important to ensure the reliability of the source data, the protection of confidentiality and the suitability of the data used for the assigned tasks.

In Japan, the basic principles of using artificial intelligence for participants in the pharmaceutical industry were presented: the document defines approaches to the use of neural networks in the development and evaluation of drugs The Center for Regulatory Science of the Japanese Medicines and Medical Devices Agency (PMDA) published a guide, "Guidelines for the use of artificial intelligence in regulatory activities (preliminary consideration)", which sets out approaches to the responsible use of artificial intelligence (AI) in the pharmaceutical industry. The document covers the areas of drug development, evaluation and regulation, including the use of neural networks to support regulatory judgments. It is mainly aimed at pharmaceutical companies, sponsors of clinical trials, developers and suppliers of AI models, as well as the regulators themselves. The guide describes the possibilities of using neural networks, including generative AI, during development, preclinical and clinical studies, registration and manufacturing. The document does not impose strict requirements on participants in the pharmaceutical market - it only specifies general principles for the responsible use and development of AI. PMDA proposed a three-tier structure. The human-centered principle comes first, followed by six basic principles and five areas of application and management. * So, AI cannot replace human judgments and decisions, but only support them. It is not acceptable to rely only on the neural network - the main responsibility will remain with the person. The six core principles include: A risk-based approach - the control measures must be appropriate to the level of risk. Correct definition of the context and the scope of the application - the goals of using AI, the accuracy of the results and the limits of the application must be determined in advance. Data management - it is important to ensure the reliability of the source data, the protection of confidentiality and the suitability of the data used for the assigned tasks.