AI Tool 'KIBIT': A Game Changer in New Drug Development

このページを日本語で読む

The COVID-19 pandemic reminded the world of the importance of drug discovery. However, modern drug discoveries require massive financial investment. Meanwhile, the use of artificial intelligence for AI drug discovery is gaining attention as a way to control soaring pharmaceutical prices. And a Japanese-made AI tool named 'KIBIT' is about to open up a new stage in this field.

America's Mandate

President Donald Trump issued executive orders in April and May mandating the reduction of domestic drug prices in the United States. Accompanying White House fact sheets refer to the high cost Americans pay for prescriptions, which is 2.78 times higher than the OECD average. Prescription costs are also 3.47 times higher than in Japan. Trump sees these high costs to consumers as a serious issue.

In Japan, a national drug pricing system sets medication prices, enabling access to affordable treatments due to governmental oversight. In contrast, US pharmaceutical companies can set prices freely. This often results in higher charges than in other countries.

Following his May 12 executive order, Trump stated that his orders would cut prices by 59% to as much as 90% to align with the levels of other countries. This could potentially result in significant revenue losses for the pharmaceutical industry. In the case of Japan, major pharmaceutical companies such as Takeda Pharmaceutical Company Limited and Astellas Pharma Inc earn over 30% of their revenue from the US.

According to his comments, the President could impose additional tariffs on foreign-manufactured drugs if prices are not reduced.

*In the search for new drugs and treatments (Screenshot, ©FRONTEO Inc)*

Rising Prices Due to Increased R&D Spending

One major reason for rising drug prices is the increasing cost of research and development. The average R&D spending by Japanese pharmaceutical companies rose from ¥30.2 billion JPY ($298.7 million USD) in 1993 to 163.3 billion ($1.13 billion) in 2019 — a 5.4-fold increase over 26 years. US companies saw an even sharper rise, from $841 million (¥121.1 billion) to $7.449 billion (¥1.0725 trillion). That marked an 8.8-fold increase.

Moreover, investing in R&D doesn't guarantee success. The probability of developing a successful drug in Japan dropped from 1 in 13,000 two decades ago to 1 in 23,000 recently. Despite the increasing costs, results are scarce. Consequently, the R&D-to-revenue ratio rose from about 10% in 1993 to around 18% in 2019 in both countries.

*Introducing KIBIT. (Screenshot ©FRONTEO Inc)*

Can AI Reduce R&D Costs?

AI is being eyed as a solution to reduce R&D costs by dramatically cutting time and expenses.

Drug development involves four stages:

Basic research & target identification
Compound optimization
Preclinical trials and
Clinical trials.

Many AI vendors are involved in drug discovery. In particular, they focus on reducing costs in the third (preclinical) and fourth (clinical) stages.

However, most companies have yet to tackle the critical first stage — target identification.

Proposing New Drugs from Papers - A Breakthrough from Japan's AI 'KIBIT'

The first enterprise to address this initial stage is the Japanese company FRONTEO. Utilizing its proprietary natural language processing AI engine KIBIT, the company analyzes vast medical and pharmaceutical literature to generate innovative drug ideas that researchers may not have considered.

Human thinking is inherently biased. The more experienced a researcher is, the more difficult it becomes to identify novel molecular targets. Stated another way, the bias of experience often blocks new ideas.

KIBIT, however, is free from such biases. It identifies new molecular targets that might be effective against specific diseases. To do so, it analyzes vast amounts of academic literature. It can even derive specific hypotheses.

Furthermore, KIBIT can identify and suggest highly disease-relevant target molecules that are not explicitly mentioned in the literature. This significantly enhances drug discovery potential.

*How KIBIT works. (Screenshot, ©FRONTEO Inc)*

Toward an Era of AI-Created Marketable Drugs

Only a few countries have the capacity to develop new drugs. In 2024, among newly approved pharmaceuticals in Japan, the US, and Europe, the US led with 143 products. Japan followed with just 12, roughly equal to the United Kingdom's 10.

According to the statistics "Nationalities of companies creating the top 100 drugs in the world by sales (2022),":

United States – 52
United Kingdom – 10
Switzerland – 9
Germany – 8
Denmark – 8
Japan – 7

Most countries, other than the US, struggle to produce blockbuster drugs.

In this context, hopes are high for FRONTEO's KIBIT. If it can consistently generate innovative new drugs efficiently, it may help deliver affordable medicines to those in need, without Mr Trump's intervention.

FRONTEO's AI drug discovery service, utilizing KIBIT, is already being adopted by several major pharmaceutical companies. Those leveraging KIBIT to create new drugs could become game changers in the industry.

This article is contributed by FRONTEO Inc, a supporting member of JAPAN Forward.

(Read the report in Japanese.)

Author: FRONTEO Inc.

このページを日本語で読む