What R&D leaders can learn from AI-driven drug discovery

Author: Saskia Hoving

Editor-in-Chief

AI is increasingly used in drug discovery and development. By generating hypotheses and uncovering connections, it supports researchers in making first-in-class discoveries. The key to this ability lies in combining high-quality, structured data with systems designed to identify opportunities and collaborate with researchers to validate them. With insights from FRONTEO’s Director and CTO, Dr. Hiroyoshi Toyoshiba, we explore how AI is able to suggest novel ideas, and why human expertise still matters.

In today’s pharmaceutical landscape, the pressure to innovate is relentless. R&D teams are expected to deliver faster, cheaper, and more targeted therapies, yet the path from idea to approval remains long and uncertain. With ever-growing volumes of data and scientific literature, it is impossible for researchers to consider it all. AI is increasingly used to conduct analyses and reach insights previously inaccessible.

AI is used in various stages in drug development, but it cannot do it alone. As Dr. Hiroyoshi Toyoshiba, Director and CTO at FRONTEO, a company that developed the Drug Discovery AI Factory, puts it: “Even a large amount of data and AI will not necessarily produce the right answer. Discovery is not something that can be achieved by AI alone.”

The most promising approaches use AI to generate hypotheses and uncover non-obvious connections to support researchers in making first-in-class discoveries. “It is ultimately humans who turn those hints from AI into innovative discoveries,” says Dr. Toyoshiba. These lessons are not just relevant to FRONTEO, they’re applicable to any organisation navigating the complexities of modern drug development.

From data overload to AI-supported hypothesis generation

The volume of academic literature is growing exponentially, and with it, the challenge of identifying meaningful insights. Traditional methods of literature review are no longer sufficient, and AI is harnessed to accomplish what humans no longer can, due to overwhelming amounts of data.

Dr. Toyoshiba explains: “Academic papers are essential sources of information for generating hypotheses. That’s why I focused on natural language processing AI — to analyse scientific literature and generate new ideas.”

This shift, from manual review of literature to AI-supported hypothesis generation, represents a broader trend in the industry. But it also raises new questions: How do we ensure the AI isn’t reinforcing existing biases, so that it can reach the full potential of its analysis and find novel ideas? And how do we to validate the insights produced by AI?

The power of unbiased AI and serendipity

AI is being increasingly utilised in the compound optimisation phase of the drug discovery process. This is a later stage of the drug development process, in which scientists attempt to improve the effectiveness and safety of a drug before moving to clinical testing. But AI indeed also has immense potential in generating new scientific hypotheses.

“Due to its nature, AI tends to lean toward computational optimisation: It excels at getting closer to the correct answer in a short period of time. However, identifying entirely novel targets remains a challenge. What we should truly aim for is the discovery of innovative and highly reliable candidates that have not yet been explored by anyone in the world,” says Dr Toyoshiba.

One of the most compelling ideas from Dr. Toyoshiba’s work is the concept of “discontinuous discovery”, finding unknown relationships from known information. This requires AI systems that are unbiased and exploratory. “One of the key points in utilising AI to discover the unknown is ‘unbiasedness.’ We are especially conscious of not relying on human biases,” he says. “Another key point is how to intentionally generate serendipity, which is considered essential for discovery.”

This approach is different to the conventional use of AI as a tool for optimisation. Instead, it positions AI as a creative partner, one that can surface surprising connections and prompt new lines of inquiry.

Why human expertise still matters

Despite the growing sophistication of AI tools, discovery in life sciences remains a fundamentally human endeavour. While algorithms can surface patterns and suggest connections, it is ultimately up to researchers to validate, interpret, and act on those insights. As Dr. Toyoshiba puts it, “It’s ultimately humans who turn those hints into innovative discoveries.”

This highlights the importance of collaborative intelligence, a model where AI augments, rather than replaces, human decision-making. For R&D leaders, the most effective strategies involve integrating AI into workflows that enhance scientific judgment.

This means ensuring that AI-generated hypotheses are rigorously tested through established methods, including preclinical and clinical validation, and that the technology fits seamlessly within existing scientific and regulatory ecosystems.

The role of trusted data

Another critical factor in the success of AI-driven drug discovery is the quality of the data it relies on. AI models are only as effective as the information they’re trained on, which makes structured, peer-reviewed, and diverse scientific literature essential. As Dr. Toyoshiba emphasises, “The reliability of data sources for analysis is extremely important. In that regard, Springer Nature offers a highly trusted database of top-tier academic papers.”

Springer Nature’s research corpus spans multiple disciplines and is well-suited for cross-domain discovery. But the broader takeaway for any organisation using AI in R&D is clear: Data governance and source credibility must be a top priority. Prioritising quality over quantity ensures that AI tools generate insights that are not only novel, but also scientifically sound and actionable.

Toward specialised AI for personalised medicine

As the pharmaceutical industry advances toward precision and personalised medicine, AI must evolve alongside it. With treatments increasingly tailored to individual patients, AI systems will also need to become more specialized, designed for specific therapeutic areas, data types, or stages of the drug development pipeline.

“Just as pharmaceuticals have shifted from general-purpose drugs to more personalised treatments,” notes Dr. Toyoshiba, “AI in the drug discovery process must also shift from a general-purpose role to highly personalised functions.”

This evolution presents both challenges and opportunities. It will require new models, regulatory frameworks, and a deeper understanding of how AI can be ethically and effectively applied in targeted contexts. For organisations willing to lead this transformation, the potential to drive more precise, efficient, and impactful discoveries is significant.

How R&D teams can leverage AI effectively

As AI continues to reshape the landscape of drug discovery, R&D leaders have a unique opportunity to rethink how innovation happens. By using AI not just to optimise but to explore, organisations can unlock new hypotheses and accelerate the path to first-in-class therapies.

The key lies in combining high-quality, structured data with systems designed for collaboration, where AI surfaces possibilities and researchers apply their expertise to validate and act on them. To move forward, teams should assess where their biggest discovery bottlenecks lie, evaluate the reliability of their data sources, and explore how AI can be integrated into their scientific workflows to support human insight.

If you're working with AI or advanced analytics in R&D, access to high-quality, structured research data is essential. Springer Nature offers tools and services that help organisations make the most of scientific literature, supporting better insights, faster hypothesis generation, and more informed decision-making.

About Dr Hiroyoshi Toyoshiba

Dr Toyoshiba is Director and CTO at FRONTEO. After obtaining his PhD in mathematics in 2000, he conducted data analysis in life science research in various capacities, including at the Medical Information Department of Kyushu University Hospital, the National Institute of Environmental Health Sciences (NIEHS) in the United States, and Takeda Pharmaceutical Company, Ltd. In 2017 Dr Toyoshiba joined FRONTEO, where he develops AI algorithms specialised for the field of life science. He became FRONTEO’s CTO of Life Science AI in 2019 and an executive officer in 2021. Since 2024 he also serves as FRONTEO’s Director.

Related content

• Navigating the changing landscape of drug discovery: the role of collaboration, biologics, and data
• Exploring AI and TDM in Life Sciences, Materials Science, and Fintech
• Top 5 benefits of digital labs in life sciences R&D
• Trends and practicalities of digital transformation in life sciences R&D 
• Digital lab transition: the opportunity for life science R&D labs
• Breakthrough therapies and blockbuster drug launches that are shaping healthcare
• The rise of generative AI: Transforming the role of information professionals 
• TDM and APIs: An introduction for information professionals
• Unravelling the Power of TDM: A Transformative Journey for Researchers
• Bridging the GAP with R&D departments: Three key challenges libraries face 
• The critical role of content investment in the R&D process
• 7 critical messages Info Pros want R&D scientists to know

Don't miss the latest news & blogs, subscribe to The Link Alerts!