Akihisa Shiozaki

Last fall, the Nobel Prize in Chemistry was awarded in part to researchers at Google DeepMind. AI has made drug discovery dramatically easier. Now is probably the best time ever to start a drug discovery company. Japan, however, is a weak player on the world stage. The message of the white paper I presided over at the MHLW last year is that founders cannot succeed by themselves. They need a whole ecosystem of advisers, investors, managers and production facilities around them. Taking inspiration from baseball player Shohei Ohtani’s success in the US Major League, Japan should build an ecosystem directly connected to the rest of the world.

A successful drug discovery ecosystem cannot be built without incorporating patients’ perspectives. By listening to patients, pharma companies can identify pain points and challenges, and develop the right medicines. Of course, it is important for all stakeholders — regulators, academia and private companies, as well as patients’ associations — to work together.

Unapproved drugs can help cure or extend the lives of patients. However, Article 68 of the Japanese Pharmaceutical and Medical Device Act limits access to information on unapproved products. Refractory patients thus lack access to data as well as drugs. This dearth of information can backfire by making subjects reluctant to enroll in clinical trials.

When patients request unapproved drugs, the government response should not be excessively slow. Justifiable caution should be balanced against the awareness that with intractable diseases time is often limited. Platforms where issues can be discussed openly and calmly are crucial to the successful co-creation of innovative drugs.

This panel examined how AI is currently being used, data-collection challenges and the leveraging of real-world data. Yasushi Okuno painted a future scenario of AI reading scientific literature, generating a hypothesis, getting a robot to conduct an experiment and then publishing a paper read in turn by AI to create an “AI agent-based loop.” According to Hiroyuki Tsunoda, Chugai is aiming for something similar with MALEXA, its AI-based antibody discovery technology. Meanwhile, Hiroyoshi Toyoshiba explained how FRONTEO’s proprietary AI identifies target molecules or disease mechanisms by using discontinuous discovery to find undiscovered connections between scientific publications.

Data collection and sharing is a challenge. Even within the same company, data sets are not necessarily standardized, while pharma companies are understandably reluctant to share data with rivals. Federated learning, which involves training AI models without the data being shared, is a practical solution to the data-privacy problem.

Real-world data has issues, too. The medical data of patients belongs to patients. Why expect them to provide it for free when it will become a source of corporate profit? Finding a balance is important, argued Hidenobu Ishizaki. “We need to get patients’ acceptance and make clear the benefits that will accrue to them,” Akira Izumi added. With doctors making detailed or vague notes depending on their personality, the harmonization of hospital data is another challenge. One solution is for doctors to use structured templates.

Using AI is all the more urgent for Japan as a late starter in the new modalities. Data sharing and integration across organizations can help make Japanese drug discovery globally competitive again.

The final panel was dedicated to an exploration of concrete next steps in light of the summit’s many discussions. Japan, once a pharmaceutical powerhouse, has fallen behind globally over the past decade. How can its drug discovery ecosystem be revived?

Asuka Miyabashira emphasized the importance of making Japan a more attractive marketplace. Startups looking to launch new products are unlikely to be drawn to a market with no growth — particularly when innovative drugs may face government-controlled price reductions even during the patent period.

Government should play a more direct role to enhance innovation. Smarter public investment, Shintaro Sengoku suggested, could help more startups survive the “Valley of Death” between founding and first returns. Kazuto Yamada spoke of his company’s experiences developing regenerative medicine therapies when the field was so new that they worked hand-in-hand with regulators to ensure that both public and private interests were served. Hans Klemm called for a fully developed national pharmaceutical strategy, with objectives, KPIs and remediation plans if these are not met.

Japan must also be plugged into broader, global ecosystems, with ideas flowing both ways — not least because, as Takahiko Iwaya explained, pharmaceutical multinationals have centralized their R&D operations since the 2000s. Hitoshi Nakagama noted in this context that genetic commonalities could support a multi-stage approach, with Japan strengthening links first within Asia and then elsewhere overseas.

Some changes, like eliminating outdated regulations not suitable for new modalities, are needed urgently. Other forms of ecosystem-building will require decades of patient investment. A “grand design” for reviving Japan’s drug discovery ecosystem, to borrow moderator Kiyoshi Ando’s phrase, will likely combine both perspectives.