Japan Research ensures the security of health data

Founded 77 years ago, Mitsui & Co remains dynamic by building businesses and ecosystems with new technologies such as generative AI and confidential computing.

Digital transformation takes many forms at the Tokyo-based conglomerate with 16 business divisions. In one case it is an autonomous truck transport service, in another case it is a geoanalytics platform. Mitsui is even working with a partner at the forefront of quantum computing.

A new subsidiary, Xeureka, aims to accelerate research and development in healthcare, where it can cost more than $1 billion over a decade to bring a new drug to market.

“We build companies using new digital technologies such as AI and confidential computing,” said Katsuya Ito, project manager in Mitsui’s digital transformation group. “Most of our work is done in collaboration with technology companies – in this case NVIDIA and Fortanix,” a San Francisco-based security software company.

Looking for Big Data

Although Xeureka is only three years old, it has already demonstrated a proof of concept that addresses one of the biggest problems in drug discovery – obtaining sufficient data.

Accelerating drug discovery requires powerful AI models built with data sets larger than most pharmaceutical companies have available. Until recently, exchange between companies was unthinkable because the data often contains private patient information as well as chemical formulas that are owned by the pharmaceutical company.

This is where confidential computing comes into play, a way to process data in a protected part of a GPU or CPU that acts like a black box for a company's most important secrets.

To ensure their data remains private at all times, banks, government agencies and even advertisers are using the technology, which is backed by a consortium of some of the largest companies in the world.

A proof of concept for data protection

To confirm that confidential computing would allow its customers to share data securely, Xeureka created two imaginary companies, each with a thousand drug candidates. Each company's data set was used separately to train an AI model to predict the chemicals' toxicity levels. The data was then combined to train a similar but larger AI model.

Xeureka ran its test on NVIDIA H100 Tensor Core GPUs using security management software from Fortanix, one of the first startups to support confidential computing.

The H100 GPUs support a trusted execution environment with hardware-based engines that ensure and validate that sensitive workloads are protected while in use on the GPU without impacting performance. Fortanix software manages data exchange, encryption keys and the entire workflow.

Up to 74% higher accuracy

The results were impressive. Thanks to the use of the combined data sets, the larger model's predictions were 65-74% more accurate.

The models built using data from a single company showed instability and bias issues that were not present in the larger model, Ito said.

“NVIDIA and Fortanix’s Confidential Computing essentially mitigates privacy and security concerns while improving model accuracy, which will prove to be a win-win for the entire industry,” said Hiroki Makiguchi, CTO of Xeureka, in a Fortanix press release .

An AI supercomputing ecosystem

Now, Xeureka is exploring broad applications of this technology in drug discovery in collaboration with the community behind Tokyo-1, its GPU-accelerated AI supercomputer. Announced in February, Tokyo-1 aims to increase the efficiency of pharmaceutical companies in Japan and beyond.

Initial projects may include collaborations to predict protein structures, screen ligand-base pairs, and accelerate molecular dynamics simulations with trusted services. Tokyo-1 users can leverage large language models for chemistry, protein, DNA and RNA data formats through NVIDIA BioNeMo's microservices and drug discovery framework.

It's part of Mitsui's broader strategic growth plan to develop healthcare software and services, such as boosting Japan's $100 billion pharmaceutical industry, the world's third-largest after the United States and China.

Xeueka's services include using AI to quickly test billions of drug candidates, predict how useful molecules bind to proteins, and simulate detailed chemical behaviors.

To learn more, read about NVIDIA Confidential Computing and NVIDIA BioNeMo, an AI platform for drug discovery.