Lantern Pharma (Nasdaq: LTRN) is a clinical-stage biotechnology company that uses artificial intelligence, machine learning and genomics to develop precision oncology therapies. Its proprietary RADR® platform is designed to accelerate cancer drug development and identify the patients most likely to benefit from specific treatments.
You were studying neural networks and artificial intelligence in the late 1980s, long before either entered the mainstream. What initially drew you to the field?
It started with a fascination for how the brain worked. I wanted to create an artificial brain, so I combined biology and neurobiology with computer science and early AI. Neural networks were just emerging, but the computing power and programming languages simply weren't there, so after some exciting internships, I began my career in investment banking partly because AI was limited, and I also wasn't ready for such an ambitious intellectual stage at that time.
Now the fundamentals have caught up. You could hold all the iPhones from one family in your hand and have more computing power than existed in the entire city of Boston when I was in college. The core ideas—recursive loops, embedded memory and knowledge graphs—have endured, and now we can use them to discover new molecules, predict patient outcomes, optimise dosing and test cancer therapies. What once took a year, or even longer, now takes three months or less, and we've seen a five- to tenfold improvement in performance and depth in the last five years. We're entering a golden age in medicine as data becomes more available, computing and inference steadily cheaper and accessible, and patients gain greater agency over their own health.
As healthcare becomes more data-driven, patients are becoming active participants in their own care. In oncology, how is that materially changing the way people navigate their own treatment?
It's still early days, but we're already seeing the shift. We recently were approached by a patient with triple negative breast cancer who had failed seven prior lines of therapy over three and a half years. She researched her options, found Lantern's study and asked their clinicians about it. Some leading breast cancer researchers at a major US institution didn't know about our drug program, but their patient did. That led to FDA compassionate use clearance a few weeks ago, and she'll become the first triple negative breast cancer patient to receive our investigational drug candidate, LP-184. Imagine that multiplied by 1,000 or 50,000 patients. People will increasingly advocate for themselves rather than simply rely on already overworked clinicians and saturated clinical staff.
We're seeing the same shift in clinical trials. One of our lung cancer patients in our clinical trial, The Harmonic Trial with LP-300, asked whether wearables could monitor blood pressure, blood flow and arrhythmias during his trial to provide earlier indications of safety or toxicity. We embraced the idea, collected valuable data for the clinician, and the patient continues to do well. In the future, wearable devices will become routine because we shouldn't find out two days later that something has happened—we as sponsors and drug developers should know as early as possible alongside the clinician. I expect wearables and real-time intelligent monitoring to be absolutely game-changing for drug developers and their direct interactions and relationships with patients in the future.
Lantern recently launched Open-Medicine AI to support drug developers and researchers. What problem is the platform designed to solve?
We launched Open-Medicine AI to help drug developers and researchers, initially in one of the most challenging categories of disease - rare cancers. Rare cancers absolutely break traditional models of development and economics…but still account for around 30% of cancer diagnoses and deaths each year. Open-Medicine AI combines our proprietary knowledge and models, highly curated drug development databases, bioinformatics tools and external scientific resources to answer complex questions that traditionally took weeks and months or even longer in the course of an afternoon. Rather than simply producing text, it shows its reasoning, the tools it uses and builds a knowledge graph so researchers can understand how it reached its conclusions. Our goal is simple: to make scientists better at what they do…to give them the superpowers that AI can help us with.
The platform grew out of how we were already using AI internally. We realised we could make those capabilities available to the wider scientific community. We also built expert personas—from medicinal chemists and clinical oncologists to computational biologists and clinical trial strategists—so researchers can approach problems from multiple perspectives. As I wrote in A Symphony of Intelligence, the idea is that different AI agents can work together to tackle complex problems. What if all those experts were available 24 hours a day, seven days a week for a few dollars of computing time? This was never possible before, and I expect this capability to be wonderfully disruptive and enabling for a new generation of researchers and drug developers.
Lantern now has multiple therapies in Phase II trials. What are your priorities over the next twelve months?
We've invested around $100 million using data and AI to develop cancer medicines. We now have multiple potential targeted medicines in Phase II trials across the US, Denmark, Taiwan and Japan, built a library of preclinical drug candidates, dosed more than 100 patients, published results at ASCO, AACR and the Society for Neuro-Oncology, and haven't had any safety issues—exactly as our AI predicted.
Many of the predictions our models made two, three, and four years ago regarding clinical activity, mechanisms of action, and patient response are now proving remarkably accurate in the clinic. This is naturally very exciting for us and for our collaborators.
Success in 2026 would be a successful launch of Open-Medicine AI alongside additional data from our LP300 trial for never-smokers with the L858R mutation in our LP-300 clinical trial. So far we've seen an 87% clinical benefit rate, and if those results continue with deeper responses, they could help fast-track us towards an accelerated approval. It's also exciting to see the AI we've built proving valuable beyond our own programmes.
AI is accelerating drug discovery, but progress still depends on much more than better science. Where do today's biggest bottlenecks lie?
Funding is always a challenge, and so is patient enrollment. We need to find patients and make them more comfortable joining clinical trials, which is especially difficult for a lesser-known company. If patients are choosing between Pfizer, Novartis and Lantern, you have to raise your hand and voice very high to get noticed. And in today’s regulatory environment, you have to tread carefully on your enrollment approach…where, until your drug is approved, you're very limited in what you can say. That's why the democratisation of data and opening of information directly to patients has the potential to be both empowering and game-changing. Patients want more options and more transparency.
More broadly, biotech companies need faster, responsible ways of generating meaningful human data. Bureaucracy, IRB complexities and redundancies and slow patient recruitment remain major obstacles even as AI dramatically accelerates drug discovery.
AI is moving into every part of healthcare, from drug discovery to diagnostics and consumer health. Where do you think the biggest risks lie, and what safeguards will be most important?
It depends which part of healthcare you're talking about. In drug development, we already have a rigorous process that the vast majority of the community is working to make even more rigorous. Just because AI predicts a drug will work doesn't mean we launch it. You still have to go through animal studies, phased clinical trials, and demonstrate safety and efficacy before anything reaches patients. Those guardrails already exist and should never go away. My bigger concern is diagnostics, wearables and consumer health claims, where the same level of scrutiny doesn't always exist. You'll see a lot of snake oil there first. We already see it with some peptide and wearable companies making all kinds of claims. In the major markets for medicine itself, the guardrails are strong and evolving.
What concerns me more is that AI could accelerate a divergence in healthcare – the K curve we all keep hearing about. The most innovative medicines may become available first to those who can afford them, or in countries willing to fast-track breakthrough therapies. We're already seeing that with longevity medicine, GLP-1 therapies and experimental treatments in parts of the Middle East, Asia and South America.
Over the next decade, we'll increasingly see two systems emerge: one for people who can afford early access to breakthrough medicines, and another for those who have to wait.
AI is dramatically accelerating drug discovery, but clinical development continues to move at a much slower pace. How will biotech companies need to rethink the way they generate clinical evidence?
Biotech companies will increasingly need to consider alternative ways to generate high-quality human data. One patient won't make enough difference to accelerate approval, but if we can treat two or three patients who otherwise wouldn't have entered a trial, that starts to become meaningful. For example, we're planning a paediatric oncology trial in the United States that won't begin until the end of the year. If, during those six or seven months, a specialist paediatric oncology centre elsewhere could safely treat a small number of patients, share the data and follow the appropriate safety protocols, those are the kinds of opportunities companies will increasingly have to consider. And very importantly, that data can help inform the trial and make investors more comfortable with the risks of involvement.
We've experienced how long the current system can take. One of our trials in Denmark took around 15 months from our first discussions to approvals. Bureaucracy, IRBs and the pace of the existing system are becoming harder to reconcile with the speed of scientific discovery. Biotech companies will increasingly have to find new ways of generating meaningful data that improve drug development, support fundraising and help us better understand how medicines should be used.
