AI is changing how fast companies can discover, test, and bring cancer drugs to market – and in this conversation, we get a front-row look at what that actually looks like in the real world.
In this episode of The Brainiac Blueprint, Kyle sits down with Aditya Pai, Head of Business Development at Genialis, a precision oncology company using machine learning to make cancer treatment more accurate, more predictable, and more personal.
It’s a deeply technical mission backed by a deeply personal story.
Before we dive into the science, the conversation begins with the origin that shaped Aditya’s career.
A Personal Mission That Became a Professional Calling
Before leading business development for Genialis, Aditya went through an experience that reshaped how he thought about cancer treatment entirely.
His mother was diagnosed with non-small cell lung cancer in 2012. At the time, targeted therapies were still limited. Still, she received one – and survived 1,000 days, despite being told she would likely live no more than 100.
Years later, on the anniversary of her passing, Aditya wrote a LinkedIn blog reflecting on how fast cancer therapies had advanced in just five years. That post caught the eye of a future co-author, and together they wrote A Race Called Life, a fiction book inspired by families facing cancer where access to diagnostics and treatments can determine outcomes.
“All proceeds from that book go to Memorial Sloan Kettering,” he shared. “It was not a commercial venture by any stretch, largely one meant to inspire people.”
This combination of personal experience, scientific understanding, and mission-driven work became the throughline of Aditya’s role today: helping companies bring life-saving drugs to patients faster, more precisely, and with less guesswork.
How Genialis Thinks About Precision Oncology
Once the groundwork was set, Kyle asked Aditya to define precision oncology in plain language.
Aditya explained it this way: finding the right drug for the right patient at the right time. But Genialis doesn’t stop at matching a mutation to a medicine – their approach digs much deeper into the entire biological system of a cancer.
The conversation walks through legacy biomarkers versus next-generation biomarkers. Physicians are familiar with simple biomarkers like hemoglobin A1c or a KRAS mutation. Traditional KRAS testing, for instance, tells clinicians whether a single mutation exists – but not whether the patient will respond to a specific drug, resist treatment, or how long the response might last.
Genialis models the entire KRAS biological pathway instead of just a mutation on it. This holistic, pathway-level view enables them to build machine-learning biomarkers that answer far more complex questions.
As Aditya put it, “We are able to be much more precise. We’re able to effectively tell when a patient will benefit from that particular drug or perhaps a combination therapy that otherwise would not be achievable.”
He shares that Genialis recently presented research at ASCO showing how these models can predict when a cancer therapy is likely to stop working and what treatments might come next – a significant step toward more personalized oncology.
Inside the Genialis “Supermodel”: A Foundation Model for Cancer
Kyle then asked how much data this level of precision actually requires.
The answer: an enormous amount.
Genialis built a foundation model – what they call their “supermodel” – on 14 well-known hallmarks of cancer established by Hanahan and Weinberg. These hallmarks represent the broad biological reasons cancers develop.
From these hallmarks, Genialis created 150 biomodules trained on whole-transcriptome RNA data from diverse populations around the world. RNA, Aditya explains, shows what is actively happening in a tumor – not just what might happen based on DNA.
Each biomodule acts like a building block that can be combined in different ways depending on the drug target. Whether a biotech team is working on antibody-drug conjugates, immune checkpoint inhibitors, DNA damage repair agents, or KRAS inhibitors, Genialis can reconfigure these modules into a tailor-made biomarker.
This approach avoids the “black box” stigma of AI. As Aditya emphasized, “There’s no magic black box… We can go down to each biomodule and show exactly how a sample is performing.”
For drug developers, that transparency matters.
Why Drug Development Takes 10–15 Years – and How AI Reduces the Risk
Kyle shifts the conversation to a topic every biotech founder understands: the brutal timeline of drug development.
From early target discovery to FDA approval, the process typically lasts 10–15 years and costs around $2 billion.
With 90% to 95% of cancer drugs failing, a single misstep in phase two or three trials can bankrupt a company.
Aditya explains why an AI-driven, biomarker-first approach is so important. The earlier a biotech company can understand how its drug behaves, the more likely it is to avoid costly surprises.
For example, Genialis can build a biomarker during preclinical or phase 1 studies – long before a company invests in large patient cohorts. The biomarker can then be used as a selection tool to enroll only the patients most likely to respond.
This reduces wasted spend, improves trial success odds, and increases the likelihood that the eventual treatment will truly benefit the right patients.
It’s not just a scientific advantage – it’s an economic one.
The Business Development Side: Global Data, Strict Standards, and Collaborative Science
Kyle then asks what Aditya’s role looks like day-to-day, especially when sourcing the massive datasets needed to build accurate models.
Aditya breaks it into two worlds:
- Bringing high-quality data in.
- Partnering with pharma, biotech, and diagnostic companies to bring accurate biomarkers out.
On the data side, Genialis partners with organizations like the Pancreatic Cancer Action Network, Sidra Medicine in the Middle East, Academia Sinica in Taiwan, and Tempus AI. Each provides unique transcriptomic data or real-world evidence that strengthens the supermodel.
But data alone isn’t enough. Genialis built a rigorous normalization pipeline that turns raw RNA sequencing files into “machine-learning-ready data.”
This ensures every dataset – regardless of country, lab, or method – is comparable.
As Aditya explains, “It is all about the quality of data.”
On the output side, Genialis collaborates with drug developers to build biomarkers, and with diagnostic companies to eventually convert those biomarkers into clinical trial assays or companion diagnostics.
It’s a complex ecosystem, but one built around a simple goal: choose the right patients from day one.
The Future: Unified Cancer Testing and More Accessible Diagnostics
The conversation then moves into the future of testing.
Today, lung cancer alone can require many individual tests – each tied to different mutations, drugs, or follow-up treatments. It’s confusing for physicians and overwhelming for patients.
Aditya believes a unified test is possible. Not today, but soon.
The challenge isn’t just scientific. It’s regulatory, educational, and tied closely to reimbursement. A unified test must be easy for physicians to understand, affordable for payers, and actionable for patients – and most current tests are linked to specific drugs.
Even so, Aditya sees consolidation coming. One test, one report, one interpretation path. Less friction at every step.
AI at Home: The Next Era of Patient Empowerment
Before wrapping, Kyle asks how AI will eventually reach patients directly – not just researchers and drug makers.
Aditya believes we’re already seeing the early steps.
Generative AI gives patients access to information that once required a specialist. They can ask smarter questions, interpret reports more easily, and explore options with more clarity.
The next frontier? At-home diagnostics.
Aditya imagines a world where early detection tools are as common as smart watches – perhaps even smart toilets that detect early signs of gynecological or urological cancers.
“Why not?” he says. If patients can detect cancer earlier, outcomes improve dramatically.
This future isn’t guaranteed, but it’s plausible – and AI sits at the center of it.
Rapid Fire: A Few Closing Moments
If you could automate one workflow instantly…
“Get the best drug to the best patient at the right time.”
Your biggest ‘aha’ moment?
- “Seeing the power of immune checkpoint inhibitors… realizing treatment could be tissue-agnostic.”
- “In university, understanding how molecular biology works – the complexity explains where we are today.”
If AI could answer any question… what would you ask?
“When will we get to an 80%+ response rate for certain cancers?”
Title of your autobiography?
“The Race Is Ongoing.”
Favorite hockey player?
“Teemu Selänne – The Finnish Flash.”
Final Thoughts
Aditya closed the episode by bringing everything back to people, not technology.
“I truly feel that the role I’m in is about really helping patients in the end.”
He spoke about friends currently battling cancer, others who passed away young, and why the industry must push harder:
“Why should that happen? It just means we’re not there yet.”
His message for anyone building in healthcare or AI:
“How do we drive ourselves to becoming better and continue this pursuit?”
And the reminder that ties it all together:
“Almost all of us know someone affected by cancer. That should be the driver.”
🎧 Listen to the full conversation:
Spotify | Apple Podcasts | YouTube📄
Read the full transcript on Left Brain AI
https://www.leftbrainenterprises.biz/
