In vivo diagnostics are pharmaceuticals to diagnose or support diagnosis of diseases by directly administrating into human body including blood vessel, digestive tract, trachea and skin. A variety of agents from conventional drugs such as contrast media and radiopharmaceuticals to the latest intraoperative diagnostic drugs, which visualize tumors by fluorescent imaging technology, are categorized into in vivo diagnostics. In this review, we introduce current development and trends, and recent issues relevant to in vivo diagnostics.

It is said that the unique features of the medical device utilizing the Artificial Intelligence (AI) technology (AI medical device) are the character that specification and performance of the medical device would be changed at the post approval phrase, and also the character that performance of the medical device would be increased/decreased depending on quality of the teacher data. In these points, AI medical devices have different features from other medical devices and that would be a big challenge for the regulations of medical devices. The Ministry of Health, Labor and Welfare (MHLW) has been proceeding with preparing points to consider for the evaluation based on the report from the “Committee for promoting the utilization of AI in the field of health care”. The points to consider which was established by the MHLW on the most recent knowledge, however it is only for current AI medical devices so that it will be needed to revise the points to consider according to the development of technologies in near future. The regulatory sciences for innovative medical devices such as AI medical devices should be examined and considered in order to ensure the quality, safety and efficacy of such medical devices when used in the medical facilities with flexible manner. In this manuscript, I would like to outline the regulation of medical equipment utilizing AI technology, its current situation and current policy focusing on diagnostic imaging medical devices.

Artificial intelligence (AI)-based medical devices are currently being marketed as a tool of supporting diagnosis for medical doctors. However, there are some problems in AI to be taken into consideration in judgement of approval as a medical devices. Japanese government has initiated a working group which consists of members recommended by Academia, PMDA as well as Ministry of Health, Labor and Welfare. The mission is to make an evaluation index for AI-based medical devices. The main problems are ①unknown algorithm through the whole processing by using AI and ②change of original function of the AI-based medical devices equipped with automatic classification learning function after commercialization compared to that of the initial application. ③The data resources as well as ④data quality in the learning processing of the AI software are also very serious problems. How to deal with the problems and what are the conditions to be approved as a medical devices are being discussed in the working group. The final responsibility to use the AI-based medical devices are owed by medical doctors. Manufactures should report at the time of application about the data resources and the range of functions of the devices such as rate of accuracy or false positive rate. It is recommended that both users and manufactures better pay more attention to the risk management for usage of the AI-based medical devices.

Recent advances in artificial intelligence (AI) are propelling the development of AI-based medical and healthcare devices and systems. AI-based medical systems have new characteristics to be considered by developers and reviewers, namely, plasticity causing changes in system performance through learning, unpredictability of system behavior due to the black box nature of the AI process, and impact of advanced autonomy of AI-based medical systems on the relationship between patients and doctors. New research-and-development and medical-device-reviewing platforms need to be urgently discussed to prepare for up-coming new AI-based medical systems by considering aforementioned characteristics.

In 2018, in the computer-aided detection/diagnosis (CAD) system for medical images, it was a year commemorating the 20th anniversary of commercialization in the United States. Twenty years have passed, and especially with the development of deep learning technology in AI (artificial intelligence), CAD has evolved and diversified as AI-CAD. Along with that, the FDA has also undergone a change in the review status of CAD and its related areas. This article will touch on part of the current situation.

In 2017, National Institute of Health Sciences (NIHS) moved to King SkyFront from Yoga in Tokyo. King SkyFront is located in Tonomachi in Kawasaki and faces Haneda airport across the Tama river. It was launched in 2013 as a base for scholars, industrialists and government administrators to work together to devise real life solutions to global issues in the life sciences and environment. The NIHS has engaged in regulatory science (RS) and the four priority areas in research and study are as follows: 1) RS research supporting R&D of cutting-edge medical products, 2) RS research for foods and environmental chemical safety, 3) national test and examination and RS for health crisis management and, 4) establishment of big database of various chemicals for risk assessment and management and development of basic technology of human health risk assessment by using AI. In this report, I introduce research and study missions of NIHS, and then joint research projects of NIHS with King SkyFront members.