As part of the EMBL Entrepreneurial Minds series, organised by EMBLEM and EICAT, we welcomed Dr. James Malone to EMBL-EBI on 10 June 2026. Following the event, we spoke with him about his journey from computational biology research at EMBL-EBI to co-founding Luvida, a company using artificial intelligence to improve clinical trial design and drug development.

Dr. James Malone is Co-founder and Chief Technology Officer of Luvida. With more than 20 years of experience at the intersection of AI and life sciences, he has worked across academia, biotech, and industry, helping translate data-driven innovation into real-world applications. His career began at EMBL-EBI, where he worked in bioinformatics and data integration, long before artificial intelligence became a mainstream topic in the life sciences.

Today, through Luvida, James focuses on one of the biggest challenges in drug development: improving the design and success of clinical trials. By combining prior trial evidence, publications, behavioural and population-level data, Luvida helps clinical teams identify where a trial may struggle with recruitment, retention, adherence or diversity – and why, providing them with evidence to reduce risk and improve outcomes.

In this interview, he reflects on the evolution of AI in healthcare, the challenges of turning scientific advances into practical solutions, and the lessons he has learned from playing the “long game” in science, technology, and entrepreneurship.

Photo credit: Laurène Ramos Martins

  • Your career has taken you from early bioinformatics research to leading AI initiatives in drug discovery and clinical trials. How did that journey unfold?

My introduction to bioinformatics came at EMBL-EBI, where I worked on ontology-based data representation and projects such as the Gene Expression Atlas. Much of the work involved harmonising and cleaning biological data, making it computationally usable and easier for scientists to interpret. At the time, the focus was largely preclinical. Over the years, however, I became increasingly interested in how these datasets could be connected to clinical information and ultimately used to support decision-making in healthcare. The challenge was no longer just organising biological data, but integrating very different types of information, preclinical, clinical, lifestyle, and real-world data, to create a more complete picture of human health. That progression naturally led me towards AI and its application in drug discovery and clinical trials, where the goal remains the same: making complex data more understandable and actionable for researchers and clinicians. In many ways, I’m still working on the same problem: how do we make complex biological and clinical information computable enough to support better decisions?

  • You were applying neural networks to biological data long before AI became mainstream. What attracted you to these methods so early on?

I’ve always been fascinated by the relationship between humans and computers. Growing up, I was a huge science-fiction fan, and I was intrigued by the idea of machines understanding and interacting with people in meaningful ways.

When I encountered AI during my university studies in the early 2000s, it immediately felt like the right tool for tackling real-world problems. I was particularly interested in how computers could learn, interpret information, and make decisions in ways that resembled human reasoning.

At the same time, I wanted to work on problems that could have a tangible impact on people’s lives. Biology and healthcare offered exactly that. The idea of making computers smarter while applying that intelligence to important scientific challenges was incredibly exciting, and it still is today.

  • Looking back, what has changed most in the life sciences over the last 20 years that has made AI such a powerful tool today?

The biggest change has been the availability of high-quality data.

Institutions such as EMBL-EBI invested heavily in creating open, curated biological datasets, laying the foundations for many of today’s breakthroughs. Projects like AlphaFold would not have been possible without decades of work dedicated to generating, standardising, and sharing data.
The second major change is scale. While many of the underlying AI concepts have existed for years, today’s models can process vastly larger datasets and identify patterns at a speed that simply wasn’t possible before. Together, improved data availability and advances in computational infrastructure have transformed what AI can achieve in the life sciences.

  • You’ve worked in academia, startups, and large organizations. What lessons have you taken from each of these environments?

Academia provides the freedom to explore new ideas and tackle problems that nobody else is working on. It’s where truly novel concepts often emerge. Many of the approaches we use today originated from research environments where people had the time and space to experiment.

Startups are different. They force you to focus on solving real problems quickly and efficiently. You learn how to validate ideas, adapt rapidly, and ensure that innovation delivers tangible value.

Large organisations operate at a different scale again. They face many of the same scientific challenges, but with far more complexity and operational constraints. What I’ve learned is that the core scientific questions often remain remarkably similar across all three environments. The difference lies in how quickly you can move and the scale at which solutions need to operate.

  • AI promises to transform healthcare, but progress often feels slower than expected. Why do you think that is?

Healthcare moves carefully for good reason. Drug development directly affects patients’ lives, so every new approach must be rigorously tested and validated before it reaches patients.
The reality is that developing a new medicine can take ten to fifteen years. Even if AI significantly improves parts of that process today, it may take years before those improvements are reflected in approved treatments and patient outcomes.
That can create the impression that progress is slower than expected. In reality, AI is already having an impact. Researchers can analyse information faster, make better decisions, and reduce inefficiencies throughout drug discovery and development. The benefits are real, but many of them happen behind the scenes long before a new medicine reaches the market.
At this point, I also think the question is no longer whether AI should be part of healthcare. We’re not going backwards. AI is becoming embedded across research and development, so the focus should be on using it responsibly, transparently, and in ways that genuinely improve outcomes for patients.

  • At Luvida, you’re using AI to improve clinical trial design. How can better use of data help make trials more successful?

One of the biggest risks in clinical trials is participant recruitment and retention. Trials often struggle because patients do not enrol, or because they leave before completion.
Our focus is on understanding the hidden variables that influence those outcomes. Beyond clinical characteristics, factors such as lifestyle, behaviour, demographics, and personal circumstances can all affect whether someone participates and remains engaged.
By surfacing those factors, we can identify potential risks earlier and help trial sponsors design more patient-centred studies. Sometimes that means adjusting elements of the protocol itself; more often it means tailoring how participants are engaged throughout the trial. Ultimately, it’s about understanding people better and designing studies around real lives rather than assumptions.

  • You often talk about integrating biomedical, clinical, and real-world data. What makes this so challenging in practice?

The challenge is that these datasets were never designed to work together.

Biomedical research data, clinical records, wearable devices, and patient-generated information all use different formats, standards, and vocabularies. Before any meaningful analysis can happen, those data sources need to be unified and represented in a consistent way.
AI helps manage the scale of this problem, but there are still major challenges around data quality, representation, and statistical rigour. More fundamentally, we still face an important scientific question: how do we translate all of these signals into decisions that ultimately lead to better medicines and treatments?
We can collect more data than ever before, but identifying which signals truly matter and understanding how they should inform drug development remains an active area of research. Finding patterns is one thing; turning those insights into interventions that improve patient outcomes is another.
There are also significant gaps in available data across different populations and regions. Addressing those gaps will be essential if we want AI-driven healthcare to benefit everyone equally.

  • Having worked on knowledge graphs, multimodal AI, and large language models, which technologies are you most excited about today?

What excites me most is the emergence of AI as a companion that can support researchers, clinicians and patients in more practical, context-aware ways.
The concept itself isn’t new, but the infrastructure supporting these systems has improved dramatically. We’re moving towards AI that can genuinely assist people in their daily lives and help them make healthier and more informed decisions.
At the same time, I believe we need to invest much more in transparency and trust. As AI systems become more capable, we must think carefully about responsibility, accountability, and how much decision-making we are willing to delegate. AI should support human judgement, not replace it entirely.

  • Looking back on your journey, what has helped you keep moving forward and turn promising ideas into practical solutions, despite all the challenges?

A strong support system has been essential. I’m particularly grateful to my wife, whom I met during my time at EMBL-EBI. Having someone who understands the ups and downs of a career in science and technology has made a tremendous difference over the years.
I’ve also been fortunate to spend my career working on questions that continue to fascinate me. The problems I was interested in twenty years ago are still challenging today, but we’ve made significant progress.
Another important factor has been conversation. Talking through ideas with colleagues, challenging assumptions, and exploring different perspectives often helps clarify the path forward. Innovation is rarely a solitary process.

  • Your talk mentions playing the “long game” when translating innovation into impact. What does that mean to you?

Sometimes a good idea arrives before the world is ready for it.

Twenty years ago, AI was still a niche field and often viewed with scepticism, even within parts of the scientific community. The ideas were strong, but the infrastructure, computational power, and available data weren’t yet sufficient to realise their full potential.

Playing the long game means recognising that timing matters. Progress often depends on broader technological and societal developments, and sometimes success comes from staying committed to an idea long enough for the ecosystem around it to catch up.

  • What advice would you give to researchers and entrepreneurs who want to turn scientific or technical advances into real-world solutions?

Talk to potential users as often as possible.
It’s easy to fall in love with an idea, but that doesn’t guarantee it’s the right solution. The sooner you engage with the people who might use your technology, the faster you’ll learn whether your assumptions are correct.
It’s important to be confident in your vision while remaining humble enough to change course when necessary. Nobody has all the answers, and uncertainty is a normal part of innovation. If someone offers you a conversation, take it.

  • Looking ahead, how do you see AI changing drug development and healthcare over the next decade?

I think we’re about to see significant advances in drug discovery over the next five to ten years. AI will help us generate and analyse data faster, design more effective clinical trials, and move closer to genuinely personalised treatments.
I can imagine a future where you walk into a hospital and the healthcare system already understands much more about you, your medical history, relevant health data, and how you have responded to treatments in the past. AI-powered conversational systems could help clinicians navigate that information and identify treatment options that are better suited to you as an individual, rather than simply recommending the first option on a standard list. Of course, that only works if the data is used with appropriate consent, governance and transparency. Personalisation should not mean removing patient agency or weakening trust.
At the same time, these technologies won’t only benefit healthcare professionals. I think AI will also help democratise access to health information. In the future, people may have far greater visibility into their own health, with AI helping them interpret trends over time and make more informed decisions about their wellbeing.
That said, I don’t believe humans will disappear from the process. You still want the expert in the loop, overseeing decisions and applying clinical judgement. The goal is not to replace doctors, but to equip them with the information and tools they need to make more informed decisions for their patients.
AI has the potential to make healthcare more personalised, proactive, and accessible, but it works best when combined with human expertise.

We sincerely thank James for his inspiring talk and for taking the time to share his experiences and insights with the audience.

We also warmly thank EMBL-EBI for hosting the event, as well as all those involved in organising and supporting this seminar.