I've spent the last few weeks preparing for panel discussions about AI in healthcare — including one at HIMSS26 in Copenhagen next month. Those conversations have a way of making you examine what you actually believe versus what you just say at conferences.
And what I believe is this: the thing holding back personalised cancer treatment isn't a lack of data or a lack of intelligence. It's a lack of access. The data exists. The computational power exists. The AI exists. But most of it sits behind institutional walls, in silos that don't talk to each other, gated by processes designed for a world that moved a lot slower than this one.
So I'm going first.
What I've published
Today I've put my complete oncology data online. All of it. My genomic profile (APC frameshift, TP53 nonsense mutation, HMGB1 overexpressed at +2.86x). My full treatment protocol — the standard of care FOLFIRI + Cetuximab and the integrative stack I'm running alongside it, mapped to specific molecular targets. My chemosensitivity results from ex vivo CTC analysis. My tumour marker trajectory.
Every dosage. Every timing. Every rationale.
It's all on the Open Source Me page — structured, searchable, and machine-readable.
Why
Because I have stage 4 colorectal cancer. Incurable. Manageable, apparently, but with no end date and no guarantee of anything. I'm on cycle 5 of chemo. My CEA has just dropped for the first time in months — which is good — but I'm not naive enough to think the current protocol will work forever.
The upside of someone finding something I've missed — a trial I haven't seen, a synergy I haven't considered, a contraindication I've overlooked — massively outweighs whatever privacy I'm giving up. That's not a complicated calculation.
What inspired this
A bloke called Paul Conyngham in Sydney. His rescue dog Rosie was diagnosed with advanced mast cell cancer. Paul used AI to analyse her tumour DNA, partnered with the UNSW Ramaciotti Centre for Genomics, and designed a personalised mRNA vaccine for her. It worked. The tumours shrank.
"It raises the question — if we can do this for a dog, why aren't we rolling this out to all humans with cancer?"
— Martin Smith, Director, UNSW Ramaciotti Centre for Genomics
Different species. Same principle. Open data + AI + willingness to bypass the usual bottlenecks = outcomes that wouldn't happen otherwise.
I can't build my own mRNA vaccine (yet). But I can do the next best thing: publish everything and let the collective intelligence — computational biologists, oncologists, AI agents trawling the web — have a crack at it.
The risk conversation
People will ask about the risk. Publishing your complete medical data online isn't exactly conventional. But here's the thing — I've been talking on panels about patients taking control of their data, about AI enabling personalised medicine, about breaking down institutional barriers. At some point you either do the thing you're telling everyone else to do, or you stop talking about it.
For me, the opportunity far outweighs the risk. And I'd rather be the person who went first than the person who talked about going first.
What I'm asking for
Three specific things, spelled out on the data page:
1. Trial matching. UK/EU Phase 1/2 trials for TP53 reactivators or Wnt/beta-catenin inhibitors.
2. Synergy identification. Novel compounds targeting HMGB1 overexpression in MSS colorectal cancer.
3. Contraindication checks. Cross-reference my full integrative stack against FOLFIRI + Cetuximab. If something's wrong, I want to know today.
If you can help with any of that — or if you just want to look at the data and see what jumps out — the page is live.