A Structural Problem in Healthcare

I've spent my career making decisions under uncertainty—in private capital, in building companies, in environments where you never have complete information but you have to act anyway. My training is in statistics and risk (MSc, London School of Economics). I think in probabilities.

Healthcare guidelines are built the same way: population-level reasoning, expected value calculations, cost-benefit tradeoffs at scale. And they work—at scale. They save lives, reduce costs, and allocate resources efficiently across millions of people.

This isn't a criticism of doctors or the system. Guidelines save lives. But they're designed for populations, not for you—with your specific presentation, your specific history, your specific risk factors.

Someone has to advocate for the individual when the guidelines say "probably not." That someone is you. The question is whether you have the tools to do it.

In 2021, I was diagnosed with testicular cancer at 37. A simple blood test—cheap, widely available—could have confirmed the diagnosis months earlier. It wasn't ordered. At the population level, not running additional tests on a 37-year-old with vague symptoms is probably the "correct" protocol.

Testicular cancer is one of the most treatable cancers, and my prognosis is a full cure. But the delay meant months of additional chemotherapy that could have been avoided. I was treated at Princess Margaret Cancer Centre in Toronto—the care was extraordinary, but the delay was avoidable.

The information was in the system. A radiologist's note recommending follow-up was there—it just never reached me. These aren't unusual failures; they're how healthcare works when information is fragmented across providers, portals, and paper files.

The lesson: get a copy of every test result yourself. It's your data—you have a right to it, and now you have tools to understand it. AI can read a radiology report, flag a buried recommendation, and surface what matters. The more eyes on your results, the less likely something critical slips through.

First, I founded the Proactives team for the Ride to Conquer Cancer—a 220km cycle ride raising money for Princess Margaret. Over four years, we became the fastest-growing community team in the event, raising over $1 million for research in immunotherapy, precision medicine, and early detection.

Second, I went into the research. The gap between population guidelines and individual care is well-documented. Diagnostic errors are the most common, most costly, and most dangerous category of medical mistakes. The literature is clear on this. What's less clear is what to do about it from the patient side.

Being informed and asking the right questions changes everything. When you walk into a 15-minute appointment with your complete history organized, your trends visible, and the right questions ready—you get more out of that time than most people get in an hour. The goal isn't to replace your doctor. It's to maximize the time you have with them.

Large language models have now reached a threshold where they can perform genuine clinical reasoning—not symptom-searching, but structured analysis of actual medical records. A Stanford/Harvard/UVA study showed AI scoring 92% on complex diagnostic cases versus physicians at 74%. The tool finally exists.

That's why I founded Proactives.ai: a platform that takes your scattered medical records, structures them, and gives you the clinical context to walk into any appointment as an informed participant. It launched in 2025 and is free to use.