Long before he began exploring the possibilities of AI, pathology, and precision medicine, Luis Cano experienced healthcare from the other side. As a child oncology patient in provincial Peru, he spent years making seven-hour journeys to Lima for specialist care – an experience that would help inspire a career in medicine.
From the hospital wards of Chimbote to pathology, programming, and ultimately AI, Cano's journey has been driven by an enduring curiosity about how things work – and how they could work better. Here, he reflects on how his experiences inform his deceptively simple philosophy: start with the problem, ask the right questions, and never lose sight of the human being at the center of the solution.
You experienced serious illness as a child. How did that experience shape the way you think about medicine and patient care today?
I was around eight years old and living in Chimbote, Peru. I remember one night I was playing at home with my brother when my father called me over. He had noticed a mass near my right clavicle. My mother was working in a hospital at the time, so my parents immediately understood that it could be something serious.
Healthcare in Peru was very centralized at that time. The main hospitals, specialists, and most advanced technologies were in the capital, Lima, and we lived about seven hours away by bus. My mother would travel with me for my medical care while my father stayed behind to look after my younger brother. It was complicated for the whole family.
One thing I remember very clearly is how positive my parents tried to be. Their attitude was always: “We will fix it. We will find a solution. You are our child, and we will overcome this together.” They taught me to confront difficult things and, as much as possible, to look for a positive way forward.
That's how I became a patient at the National Cancer Institute. I remember lining up with my mother from six in the morning to get a ticket. Securing an appointment through the public system was also complicated, sometimes taking up to three or four months. My parents made a financial sacrifice so that I could access the private clinic system with shorter wait times, and I was able to start treatment within days. I was eventually operated on by Dr. Travezán, a head and neck specialist at the National Cancer Institute, and I would visit him regularly afterward for check-ups.
Going through that experience at such a young age gave me a different perspective on medicine. I knew what it meant to be the patient and family trying to navigate the healthcare system. I think it gave me a sensitivity to those experiences that has stayed with me throughout my career.
Did your experience as a patient influence your decision to become a physician, and what shaped you during medical school?
That experience was one of the reasons I decided to become a physician. I like to think that destiny sometimes has something prepared for you, although perhaps not in the way you expect.
I went to medical school in Chimbote, rather than Lima, which had three small- to medium-sized hospitals rather than one large national center. I spent a great deal of my medical training in those hospitals, and that exposed me to many different cases and perspectives.
I was fortunate to meet a very good teacher, Dr. Oswaldo Garcia, who told me that if I wanted more practical experience, I could join the clinical teams on their night shifts. I took that opportunity and started working nights from my fourth year of medical school.
At first, everyone thought I would become an emergency physician. I loved the intensity of that work – managing patients with heart attacks, respiratory failure, kidney failure, and other acute conditions. I learned to interpret arterial blood gases, manage ventilators, and deal with emergencies. It was excellent training, even though I ultimately took a very different path and became a pathologist.
Most importantly, it was in those hospitals that I began to understand the responsibility of being a physician and what it means to be able to help another person.
What drew you to pathology?
I really love to read. My wife is always telling me, “Another book? Please, not another book. We don’t have space for another one!”
When I discovered pathology, I realized it suited that curiosity. Pathology allows you to go deeply into how the body works and, importantly, what happens when disease disrupts that balance. To make a diagnosis and ultimately help guide treatment, you need to understand what is happening beneath the surface. That's the part I find almost addictive about pathology: you're looking at a pattern under the microscope, and behind that pattern there's a story of what went wrong in the cell, the tissue, the patient. Figuring out that story is what keeps me endlessly curious. Pathology never runs out of questions to ask.
How did that curiosity lead you from traditional pathology into computing and AI?
My interest in computers actually started long before my career in pathology. As a teenager, I taught myself how to create a website using HTML and Dreamweaver. Later, I experimented with Macromedia Flash and other software, even creating animations and tutorials to help me understand how proteins interact with one another. I have always been interested in discovering how things work and then trying things for myself.
As I went deeper into pathology, I began learning programming languages, starting with R and then some Python. That opened an entirely different perspective for me. When I discovered tools such as QuPath, I thought, “This is amazing. We can actually do this.”
Suddenly, I could start testing my own hypotheses. Why does one pathological pattern look different from another? Can we quantify that difference? Does what I am seeing visually also have a mathematical meaning? I am not a mathematician, but I began to realize how much more there was to understand about pathology if we combined our traditional methods with programming, statistics, mathematics, and AI.
Your career naturally transitioned into translational science. What principles guide you when you are trying to solve a problem with AI or other computational tools?
First, interpretation is inherently subjective. Each of us brings our own experiences and ideas, and with them, our own biases. We can't eliminate our biases, but we can be aware of them. That's why, before doing anything else, I start by asking open questions: what exactly are we trying to solve, and why? If both questions have resonance and a convincing answer, I move forward. If not, I don't, because I've learned that no matter how much technical rigor you bring, in the end it won't fix a bad question.
Getting the question right is the one place where you can actually correct for your own bias; everything downstream inherits whatever blind spot you started with.
The second is never to forget that you are working with humans, not machines, even though today so much of the conversation is about machines. Computers can be incredibly powerful at doing exactly what we tell them to do. But that is also their limitation: you have to be able to express exactly what you want in a way the machine can understand.
Humans can do something different. If I ask you to create an image of a protein, for example, you might ask: “How do you want the protein oriented? To the right or the left? What color should it be? Do you want a sense of depth?” We can be curious about what another person actually needs. If we bring that human curiosity to understanding other people’s problems, I think we have a much better chance of creating something genuinely useful.
The third principle is to always try to do your best, even when you don’t get a good result, because that is how you learn. Sometimes people think, “If I do my best, I should get a good result.” But life isn’t linear. You can do everything you can and still fail. What matters is what you learn from that experience – and from what you do differently the next time.
In your consultancy work, how does your combined experience in pathology, informatics, and AI influence the way you approach problems with your clients?
I think it is important to go deep, because that is how we develop expertise. But at the same time, we always need to step back and see the whole picture. That is how you move from creating a good product to creating something that can actually help people solve a problem.
Having different perspectives has helped me enormously. As a pathologist, I might look closely at one aspect of a disease and ask what it can tell us about the underlying biology. My experience in informatics then encourages me to ask different questions. Is the relationship we are seeing simply a correlation, or is there evidence of causality? If we think there is a causal relationship, where does that lead us?
But there is another question that I think is even more important: is this clinically relevant? You can discover something scientifically interesting and develop something technically impressive, but if it does not change a clinical decision, you have to ask why you are pursuing it.
That broader perspective is what I try to bring to the companies I work with. Whenever I look at a project, I ask the same question: if this works perfectly on paper, does it actually change what a clinician does on a Monday morning? If the answer is no, we're solving the wrong problem – no matter how elegant the pathology, the data, or the technology behind it.
Where do you see the greatest opportunity for personalized medicine to use AI more effectively?
I think there is a real opportunity for pharma to approach AI differently. Rather than starting by asking, “Which technology should we buy?”, we should start with the right question: where is our bottleneck, and how can we solve it?
It is not simply about creating more and more molecules. Of course, developing new molecules is important, but we also need to make sure that we are getting better at selecting and classifying the patients who might benefit from them.
If we develop a drug for a very broad population, we may see limited benefit overall. But perhaps within that population there is a particular group of patients who respond extremely well. If we can identify those patients more precisely, both the drug and the diagnostic strategy become much more valuable.
For me, that is what personalized medicine is about. We need to keep going deeper into the biology and become increasingly precise in how we classify patients. AI can help us do that, but the starting point should always be the clinical problem we are trying to solve, not the technology itself.
Newsletters
Receive the latest pathologist news, personalities, education, and career development – weekly to your inbox.
