Japan is developing AI-powered drug design, healthcare robotics and digital healthcare platforms

To provide high-quality medical care to its population – about 30% of whom are 65 years old or older – Japan is pursuing sovereign AI initiatives that support almost every aspect of healthcare.

AI tools based on country-specific data and local computing infrastructure are boosting the skills of Japan's clinicians and researchers so they can care for patients, despite a shortage of nearly 500,000 healthcare workers expected by next year.

The NVIDIA AI Summit Japan, taking place in Tokyo through November 13, will highlight breakthrough technology deployments from the country's leading healthcare companies, including AI-accelerated drug discovery, genomic medicine, healthcare imaging and robotics.

Based on NVIDIA AI computing platforms such as the Tokyo-1 NVIDIA DGX supercomputer, these applications have been developed using domain-specific platforms such as NVIDIA BioNeMo for drug discovery, NVIDIA MONAI for medical imaging, NVIDIA Parabricks for genomics, and NVIDIA Holoscan for healthcare robotics.

AI factories for drug discovery deepen understanding, accuracy and speed

NVIDIA supports the Japanese pharmaceutical market – one of the three largest in the world – with NVIDIA BioNeMo, an end-to-end platform that enables drug discovery researchers to develop and deploy AI models to generate biological intelligence from biomolecular data.

BioNeMo includes a customizable, modular programming framework and NVIDIA NIM microservices for optimized AI inference. New models include AlphaFold2, which predicts the 3D structure of a protein based on its amino acid sequence; DiffDock, which predicts the 3D structure of a molecule that interacts with a protein; and RFdiffusion, which designs novel protein structures that are likely to bind to a target molecule.

The platform also features BioNeMo Blueprints, a catalog of customizable reference AI workflows that help developers scale biomolecular AI models to enterprise-class applications.

The NIM microservice for AlphaFold2 now integrates MMSeqs2-GPU, an evolutionary information retrieval tool that accelerates the traditional AlphaFold2 pipeline by 5x. Led by researchers from Seoul National University, Johannes Gutenberg University Mainz and NVIDIA, this integration enables protein structure prediction in 8 minutes instead of 40 minutes.

At AI Summit Japan, TetraScience, a company developing AI-native scientific datasets, announced a collaboration with NVIDIA to industrialize the production of AI scientific use cases and accelerate and improve workflows across the life sciences value chain.

For example, selecting an optimal cell line to produce biological therapies such as vaccines and monoclonal antibodies is an important but time-consuming step. TetraScience's new Lead Clone Assistant leverages BioNeMo tools, including the NVIDIA VISTA-2D base model for cell segmentation and the Geneformer model for gene expression analysis, to reduce lead clone selection to hours rather than weeks.

Tokyo-based Astellas Pharma uses BioNeMo biomolecular AI models such as ESM-1nv, ESM-2nv and DNABERT to accelerate biologics research. Its AI models are used to generate novel molecular structures, predict how those molecules bind to target proteins, and optimize them so that they bind to those target proteins more effectively.

Using the BioNeMo framework, Astellas has accelerated the creation of chemical molecules by more than 30 times. The company plans to leverage BioNeMo's NIM microservices to further advance its work.

Japan's pharmaceutical companies and research institutions are driving drug research and development

Astellas, Daiichi-Sankyo and Ono Pharmaceutical are leading Japanese pharmaceutical companies that use the Tokyo-1 system, an NVIDIA DGX AI supercomputer developed in collaboration with Xeureka, a subsidiary of Japanese conglomerate Mitsui & Co Develop AI models for drug research. Xeureka uses Tokyo-1 to accelerate the development of AI models and molecular simulations.

Xeureka is also leveraging NVIDIA H100 Tensor Core GPUs to explore the application of confidential computing to improve pharmaceutical companies' ability to collaborate on training large AI models while protecting proprietary data sets.

To further support disease and precision medicine research, genomic researchers across Japan have adopted the NVIDIA Parabricks software suite to accelerate secondary analysis of DNA and RNA data.

Among them is the University of Tokyo Human Genome Center, the main academic institution working on a government-led whole-genome project focused on cancer research. The initiative will help researchers identify gene variants unique to the Japanese population and support the development of precision therapeutics.

The Genome Center is also exploring the use of Giraffe, a tool now available through Parabricks v4.4 that allows researchers to map genome sequences to a pangenome, a reference genome representing different populations.

AI scanners and oscilloscopes give radiologists and surgeons real-time superpowers

Japan's healthcare innovators are building AI-powered systems to support radiologists and surgeons.

Fujifilm has partnered with NVIDIA to develop an AI application designed to help surgeons perform operations more efficiently.

This application uses an AI model developed with NVIDIA DGX systems to convert CT images into 3D simulations to support surgical procedures.

Olympus recently worked with NVIDIA and telecommunications company NTT to demonstrate how cloud-connected endoscopes can efficiently run imaging and AI applications in real-time. The endoscopes were equipped with NVIDIA Jetson Orin modules for edge computing and connected to a cloud server via the NTT communications platform IOWN All-Photonics Network, which introduces photonics-based technology across the network to achieve lower power consumption, greater capacity and a to enable lower latency.

NVIDIA also supports real-time AI-powered robotic systems for radiology and surgery in Japan with Holoscan, a sensor processing platform that streamlines AI model and application development for real-time insights. Holoscan includes a catalog of AI reference workflows for applications such as endoscopy and ultrasound analysis.

A neurosurgeon at Showa University, a medical school with multiple campuses across Japan, has used Holoscan and the NVIDIA IGX industrial-grade Edge AI platform to develop a surgical microscopy application that captures video from surgical scopes and converts it into real-world 3D -Images converts time with AI. With access to 3D reconstructions, surgeons can more easily locate tumors and key structures in the brain, improving the efficiency of procedures.

Japanese surgical AI companies, including AI Medical Service (AIM), Anaut, iMed Technologies and Jmees, are exploring the use of Holoscan to support applications that provide diagnostic support to endoscopists and surgeons. These applications could detect anatomical structures such as organs in real time, with the potential to reduce risks of injury, identify conditions such as gastrointestinal cancers and brain hemorrhages, and provide instant insights to help doctors prepare and perform surgeries.

Scaling healthcare with digital health agents

Older adults are more likely to suffer from chronic diseases and use the most healthcare services. To keep up with the aging population, Japan-based companies are pioneering the development of digital health systems to improve patient care.

Fujifilm has launched NURA, a group of health screening centers with AI-powered medical exams to help doctors test for cancer and chronic diseases with faster exams and lower doses of radiation in CT scans.

Developed with NVIDIA DGX systems, the tool includes large language models that produce text summaries of medical images. The AI ​​models run on NVIDIA RTX GPUs for inference. Fujifilm is also evaluating the use of the MONAI, NeMo and NIM microservices.

To learn more about NVIDIA's collaboration with the Japanese healthcare ecosystem, watch the on-demand NVIDIA AI Summit session from Kimberly Powell, the company's vice president of healthcare.