The Missing Data in Cancer Drug Development?

Series introduction — Cancer AI: From Evidence to Models

3-minute read

Why it matters to me.

I’ve lost relatives to cancer, and more recently a friend. Each loss brought me back to the same question: why does a treatment work for one person and not another?

My years at Roche and Illumina showed me how far cancer research has come, but also how much uncertainty remains. We know increasingly more about the molecular characteristics of tumors, yet still struggle to predict how they will respond to treatment.

Knowing a cancer is not the same as knowing its response - the missing gap.

Precision medicine has benefitted enormously from genetic data, but I feel we still need more evidence connecting the molecular characteristics of a cancer with its response to treatments. AI offers extraordinary opportunities, but even the best models are limited by their training data. Much of today’s cancer data describes what tumors are, rather than how they respond to drugs. Better algorithms alone won’t close that gap.

We don't know what we don't know - a lab-in-the-loop platform lets us improve.

At Pathway Bio, we’re building a lab-in-the-loop AI platform that combines computational prediction with experimental evidence from human cancer.

Incorporating Mayo Clinic's exvivo platform (in-licensed) - patient-derived 3D tumor models, or microcancers, we measure drug responses and pair them with molecular profiles. These observations contribute to our growing Human Response Atlas, combining existing research with new experimental data.

But the opportunity isn’t simply to build a larger dataset. It’s to identify what our models don’t yet understand, design experiments to close those gaps, and demonstrate that the new evidence improves predictions.

Our working principle: predict, validate, improve, repeat.

We’re applying this approach with research and industry partners to investigate resistance, identify potential biomarkers and prioritize promising therapies. Each program addresses an immediate development question while potentially improving predictions across future programs.

I believe the next generation of cancer AI will be defined not just by better predictive models, but by the ability to generate the biological evidence those models need.

Our ambition is to turn that evidence into better decisions in cancer drug development—and ultimately better treatments for patients.

Losing people I cared about made this personal. Pathway Bio is how we are acting on it: building a deeper understanding of why drugs work or fail, so that understanding can help direct better treatments.

About this series

Cancer AI: From Evidence to Models is a short series for pharma R&D teams. A lab in the loop depends on two things: evidence that means what it appears to mean, and tests that show what a model can and cannot do. In the pieces that follow, our technical team explains:

• why public drug-response data isn’t automatically model-ready, starting with cell lines;
• the hidden work behind every cohort of patient-derived models;
• why a model is only as credible as its test; and
• what a model trained on cell lines must prove in patient-derived models.

Together, the series will show why this evidence gap matters to pharma R&D, why we believe closing it requires an integrated lab-and-model approach, how our team is building that system, and how work on immediate development questions can strengthen a broader evidence platform over time. We will remain clear about what the evidence establishes today and what we are working to prove next.

Imagine testing a cancer treatment on a patient’s tumor before giving it to the patient.

We grow living tumors from patient samples, test drug responses and use the results to teach our AI which treatments could work - and for whom.

Test the tumor. Predict response. Design smarter trials.

Connect with us at contact@pwbio.ai.

‍