This episode of ByteSight features Dr. Vinod Gauba, a physician entrepreneur who has spent over 25 years moving between the bedside and the lab. In this conversation, Manasi and Vinod explore what Vinod calls the invisible patient, the person whose biology was never reflected in the data guiding their own treatment. They discuss the deeply personal moment that reframed his entire career, why representation bias is far broader than ethnicity alone, and why fixing this problem requires more than better datasets. It requires rebuilding trust from the ground up.
About Dr. Vinod Gauba
Vinod Gauba is a physician entrepreneur with over 25 years of experience spanning clinical surgery, research, and health technology. He is the Founder and CEO of GeneVault, a Public Benefit Corporation advancing the responsible use of genomic, multi omic, and real world health data through federated collaboration and ethical governance models that keep data under local control. He also serves as Chairperson of UltraPerson, a precision prevention platform based in Abu Dhabi, and as an Advisory Board Member at Amgen for the Middle East region. He holds an MBBS from King’s College London and an MSc in Health Research from the University of Oxford in partnership with the University of Leeds, alongside surgical fellowships in oculoplastic and refractive surgery. He continues to see patients today, by choice, to stay close to the problem he is trying to solve.
From Clinician to Data Builder: A Personal Reckoning
Vinod’s shift from clinical medicine to data infrastructure did not begin as a strategic decision. It began with his own father’s cancer diagnosis. When the recommended precision oncology treatments failed to work as expected, Vinod went looking for why, and what he found reshaped the rest of his career: his father’s biology had simply never been represented in the assays, reference ranges, and clinical reports guiding his care.
His biology was never reflected in any of those assays, in any of those reference ranges, in any of those reports and results that were supposed to be guiding us and our oncologist what to do.
That experience exposed something Vinod sees as systemic rather than rare. Clinicians across the world, particularly in the Global South, have learned to quietly compensate, adjusting reference ranges, dosages, and protocols based on their own trial and error, simply because the data in front of them was never built to reflect the patient sitting across from them.
Representation Bias is Bigger Than People Think
Vinod is careful to push back on the assumption that representation bias is purely a question of ethnicity. It shows up just as urgently in gender, age, and rare disease populations, anywhere the underlying data was built around a narrow, convenient slice of humanity rather than the full range of people clinicians actually treat.
It’s actually as simple as male-female gender bias. You know, we all recognize that the majority of data feeding the literature today is predominantly male.
The conversation turns to what Vinod calls the most dangerous category of all: unmeasured bias, the kind that never gets quantified, and is therefore acted on as if it were truth. Because biopharma incentives are not structured to surface it, fixing the problem requires going far upstream, to the data itself, rather than waiting for downstream symptoms to force the issue.
What’s really scary are the biases that we can’t measure. An unmeasured bias that’s invisible, we act on it as truth, and calibrate our confidence on it. That’s the invisible patient that we’re talking about.
Trust, Not Just Data, is the Real Infrastructure
For Vinod, the deepest harm in this story is not missing data. It is misplaced trust at scale, patients trusting numbers that were never built for them, and clinicians trusting protocols they have no way to question in the time they have. He argues that any real solution has to be built with trust and transparency engineered into its foundations from day one, not bolted on afterward as a compliance afterthought.
Any solution that we build to address this problem must have incentives aligned and trust built into the foundations of how we address it.
He also points to why this matters beyond ethics: a healthcare system built on unrepresentative data simply cannot scale responsibly. Programs like the NIH’s All of Us and Our Future Health are meaningful steps toward broader representation, but Vinod is clear that good intentions are not the same as sufficient coverage, especially once epigenetic and population specific factors are accounted for.
Final Takeaway
Vinod’s vision for the next decade is simple to state and difficult to achieve: clinicians who never have to silently adjust, who can trust that the data in front of them genuinely reflects the patient in front of them. Getting there will take more than better algorithms or bigger datasets. It will take collaboration across companies, institutions, and borders, and a willingness to treat the public good dimension of this work as seriously as its commercial one.
The invisible patient, Vinod notes, is often invisible even to themselves, unaware that the numbers guiding their care were never built with them in mind. Making that patient visible again is, in his words, simply too big and too important a problem to leave to chance.
