Open invitationA hackathon on my case. Wed 25 November, 5–6pm UK, online. Oncologists, radiologists, genomics people, vaccine researchers — come and pull my data apart.Details & RSVP →
Changelog

Every update, in order

I rebuild my profile every two weeks, in time with my chemo cycle. This is the running record of what changed each time: my condition, what I worked out, what became available to download or request, and anything new that came out of putting it all in public. Newest first.

  10 updates so far · latest 17 November 2026

17 November 2026

Latest

The scans themselves, and what people say

What's new here
  • The scan images are on the site, not just the reports. Both CT scans, every slice, in both windows — the 23 July scan that found the progression, and the 28 October one that showed it responding. Scroll them with the slider, the wheel, a finger or the arrow keys.
  • Every frame of all four series was checked for burned-in identifiers — name, date of birth, hospital number — before any of it went up. There were none.
  • They are web copies, compressed to load. The original DICOM studies are still yours for the asking: the full study through the hospital's image-sharing system, or an anonymised copy on a private link.
  • It is not a diagnostic viewer, and the slice numbers are positions in a file rather than clinical references. The radiologists' own words remain the record.
What people say
  • Added what people say about FC:AI, starting with my HelloAI Patient co-authors: Jan Beger, Alvina Nadeem and Ellyn Winters-Robinson. All three gave permission.
  • The quotes live in one file and the pages render from it, so the site cannot drift from what somebody actually approved — the checker compares every published quote back, character for character, and fails the build on any difference.
  • The home page note about HelloAI Patient is now a proper thank you rather than a one-line announcement.
  • Ellyn's surname is hyphenated, Winters-Robinson. It was wrong in two places for a day. Fixed, and the checker now catches it.

16 November 2026

A course for patients, and a tools page that grew up

What's new here
  • HelloAI Patient launched, and I co-wrote it. A free course on using AI well with your health, for patients and the people who care for them: four short courses, 23 units, under an hour. Written with Ellyn Winters-Robinson and Alvina Nadeem, led by Jan Beger, from the HelloAI programme led by GE HealthCare. The post.
  • The tools page is now Tools for patients: a map of the tools I've made, the ones I trust and the ones I'm watching. Four tiers, the same labelled fields on every entry, a check date on each one, and sources at the bottom.
  • Anything I haven't tried is marked as such, and what's written about it is what the tool says about itself, not my verdict.
  • The URL hasn't changed. fcancerwith.ai/tools still works, because the course links to it and so do older posts.
What I checked
  • The course page says four courses and 23 units, under an hour. That's the figure on the page here too, rather than my own rough “about 40 minutes”.
  • opencancer.ai's own words are “no pharma sponsorships, no selling your data”. That's what the page now says it says.

13 November 2026

Cycle 7, and a CEA that went the wrong way · 29 Oct – 13 Nov

My condition
  • Dropped the irinotecan at cycle 7, on 12 November. First maintenance cycle: modified de Gramont and cetuximab. The cold cap went too, and the erythromycin, and the steroids are down to the two days after the infusion. It should make the chemo a good deal easier.
  • CEA more than doubled to 104.0 on 11 November, from 46.5 on 28 October. It had sat at 44 to 47 since late September. Which is less fun.
  • ALP is up too, 149 to 166. Everything else is in range, bar red cells at 4.31. White cells and neutrophils have risen but are still inside range.
  • The CT two weeks earlier showed a partial response, so the scan and the marker now disagree. Nothing changes on one blood test: repeat bloods on 25 November, then the telephone review with a CT on 8 December.
  • Written up in full: Dropped the irinotecan. My CEA didn't get the memo.
What's new here
  • One View is now v1.6. The 11 November panel is on the card, the page, the full record and the data feeds, and the 11 November lab report is published with identifiers redacted.
  • The CEA chart on the full record now runs to 23 readings and shows the spike.
  • The wallet card has irinotecan marked as stopped, so anyone checking what I'm on gets the current answer.
  • The checker now fails the build if any current-status text still calls the CEA steady or flat. It was true a fortnight ago and isn't now.

4 November 2026

The letter and the reports · 3 – 4 Nov

My condition
  • The clinic letter has arrived and is summarised in full on My One View. The formal wording: omit irinotecan from cycle 7, continue maintenance modified de Gramont and cetuximab to disease progression.
  • Local treatment is not indicated. My symptoms improved on chemotherapy, and further ablation or radiotherapy would risk worsening my breathing with no impact on survival. The liver responded too, so the same applies there.
  • At progression, irinotecan could be added back. After that, Lonsurf and bevacizumab, then clinical trials, then fruquintinib.
  • My smouldering myeloma may exclude me from some trials. That gets checked against each trial's criteria nearer the time.
  • Next review: telephone before cycle 9, 8 December, with CT results.
  • Metformin, mebendazole and atorvastatin are now disclosed and confirmed as safe alongside the chemotherapy.
What's new here
  • The written radiology reports are published for the 28 October and 23 July scans, identifiers redacted and the findings untouched. Worth knowing: the October report gives no lesion measurements. The July one does.
  • I hold the complete original scan images — the DICOM studies, not just the reports. They aren't published, but clinicians can have the original through the hospital's image-sharing system and reviewers can have an anonymised copy on a private link.
  • There is now an ask me for anything section on One View. The images, the raw sequencing files, the source document behind any figure, the record in another format. You only have to ask.
  • Two older PDFs were re-redacted: they still carried my date of birth in a format the old check missed. The check now matches patterns rather than spellings.

3 November 2026

The results call · 2 – 3 Nov

My condition
  • The cancer is responding, so I am dropping the irinotecan. At the results call on 2 November we agreed to move to 5-FU/folinic acid + cetuximab as maintenance from Cycle 7, around 12 November.
  • My oncologist reads the 28 October CT as a response at all the chest sites, with the liver possibly a little smaller and nothing new. A true response, not just stable disease. Each rechallenge gives a slightly smaller response, which fits the CEA settling around 46.
  • The aim of maintenance is disease control and keeping the cough away. More shrinkage is not necessarily expected, and that is fine. Cycle 6 was given on 29 October. A short Christmas break is agreed in principle.
  • Local treatment to the lung or liver is not advised now: too many nodes and nodules, and I have had lung radiotherapy before. The liver lesion gets revisited at the next scan.
What comes after this
  • The agreed sequence is Lonsurf (trifluridine/tipiracil) + bevacizumab, then clinical trials, then fruquintinib or regorafenib. Oxaliplatin is ruled out — I progressed on it in 2022 and had a severe reaction.
  • Whole genome sequencing on tissue is not funded and would need fresh frozen tissue, so an NGS blood test is proposed instead. Funding is being checked.
  • One View is now v1.4. The clinic letter and the formal radiology report with measurements are still to come, and will be published when they arrive.
  • The hackathon is Wednesday 25 November, 5–6pm UK. This plan is what it will be built around.

2 November 2026

The restaging CT · 28 Oct – 2 Nov

My condition
  • The CT on 28 October shows a partial response. The lung metastases and intrathoracic lymph nodes have partially reduced, including the subcarinal and left AP window nodes; a right upper lobe nodule has cavitated. The segment VII liver lesion is potentially minimally reduced and overall stable. No peritoneal disease.
  • The radiologist's comment, in full: “Partial response of the intrathoracic lymph nodes and lung metastases. Stable appearances of the liver.”
  • So the CEA was flat at 46.5 while the scan improved. That is worth knowing about tumour markers: mine has been a poor guide to what the imaging shows.
  • At this point nothing was decided. The questions were: treat the liver lesion? Continue, or drop the irinotecan, given UGT1A1 *28/*28? What is the next line? The answers came the next day.
What's new here
  • One View is now v1.3, redesigned with a front and a back: the front is the one-minute version, the back is the detail a clinician asks for. Both print one page per side.
  • A CT head from 19 June 2026 is on the record for the first time. It was normal — no lesions, no leptomeningeal enhancement.
  • The imaging history is now structured data in oneview.json, so it can be read by other tools rather than only by people.
  • The hackathon is still Wednesday 25 November, 5–6pm UK. The scan result is what it will be built around.

29 October 2026

Cycle 5 · 12 – 29 Oct

My condition
  • CEA is 46.5 µg/L on 28 October, against 47.3 on 12 October. Down 68% from the 1 September peak of 146, and steady since late September — flat rather than falling or rising.
  • ALP has come down for the first time in months: 190 on 12 October to 149 now. Haemoglobin is back inside range at 134, bilirubin is 9, and platelets, white cells and neutrophils are all normal. Red cells remain marginally low at 4.31.
  • Cycle 5 of the FOLFIRI + cetuximab rechallenge was given on 14 October. The restaging CT is done and the report is due the week of 2 November.
What's new here
  • My pharmacogenetics are published: I am UGT1A1 *28/*28, a poor metaboliser — slower irinotecan clearance, higher risk of neutropenia and diarrhoea, and the reason for my Gilbert's.
  • A lab cell-kill test on circulating tumour cells is published, with the caveat it deserves: a dish is not a person, and it is no proof of benefit in me.
  • A hackathon on my case, Wednesday 25 November, 5–6pm UK, online. Details and RSVP.
  • The 29 September platelet count is recorded at 280 on the full record, which had been missing.

21 October 2026

Cycles 4 and 5 · 1 – 21 Oct

My condition
  • Open Source Me is now My One View. Same data, same openness, clearer name. The page lives at fcancerwith.ai/my-one-view and the sheet at /my-one-view/card. Every old link redirects.
  • CEA is 47.3 µg/L. 146 on 1 September → 47.3 on 12 October; steady at about 45–50 over the last three draws. The hard fall across the restart has levelled off rather than continued, and one reading is not a trend — the restaging CT is what answers it. ALP has gone 102 (July) → 165 (1 Sep) → 179 (14 Sep) → 169 (29 Sep) → 190 now. Haemoglobin and red cells are still just under range, bilirubin is back inside it at 14, and red cell distribution width is flagged high for the first time. The 29 September platelet count is confirmed at 280, inside range.

1 October 2026

Cycles 3 and 4 · 17 Sep – 1 Oct

My condition
  • CEA is 44.6, down from 50.9 on 14 September and 146.0 on 1 September. Third fall in a row — about 69% across three cycles of the restart.
  • Cycle 3 was physically easier than Cycle 2. Same drugs, same doses; what changed was soup-only through the nadir weekend and stepping the laxatives down once things moved. No GI emergency this time.
  • Haemoglobin (126) and red cells (4.09) have drifted just under range. ALP is still high at 169. Bilirubin 21, marginally over — I have Gilbert's syndrome.
  • Cycle 4 started on 30 September. Cycle 5 is 14 October, with a CT scan after it.
What I found out
  • I finally pinned down what sequencing I've actually had: a 517-gene panel, aligned to hg19. Not whole genome. Not whole exome. A lot of people assumed otherwise, including me.
  • The tumour panel is reported tumour-only, so my inherited variants are sitting inside it — that matters for anyone reanalysing it.
  • Daily tracking now runs to 29 September: 40 days, morning and evening, across three cycles.
  • The cycle map is now Cycle 4, with metformin established from day 4.
New to download or request
  • The full manifest of my raw sequencing package — 17 files, 36GB, every size and MD5 checksum. The files themselves are available on request.
  • Daily tracking as a CSV, 20 August to 29 September.
  • The 29 September bloods as a PDF, identifiers redacted.
  • The Cycle 4 map as a PDF.
  • The One View as PDF, plain text and JSON.
What it flagged up
  • A genomics researcher read this site and pointed out that panel testing isn't whole genome sequencing — and that sequencing tumour and normal tissue together is what opens a personalised therapy route. Three people have now said the same thing independently.
  • There's an NHS route to it. The South West Genomic Medicine Service covers Bath. That's the next call I make.
  • An oncologist has offered to work on vaccine design voluntarily, and flagged that sequencing deep enough for that is a different, deeper job than clinical sequencing.
  • Five vials of blood went to a blood analysis company at the end of September.
  • One gap I can't close yet: the platelet figure on my copy of the 29 September report is unreadable, so it's blank rather than guessed.

17 September 2026

Cycles 1 to 3 · 19 Aug – 17 Sep

My condition
  • CEA 50.9 on 14 September, down from 146.0 on 1 September — the first big fall after restarting chemo in August.
  • ALP high at 179, globulin 37, platelets 289. Bilirubin normal.
  • Cycle 3 infusion on 16 September. Chemo restarted 19 August after progression during the summer break.
What I found out
  • Rebuilt the whole site around Open Source Me (now My One View): one card carrying my entire profile, always the latest version at the same URL.
  • Split the full record into Current and Archive, so what matters now is at the top and the older material is still there, just further down.
  • Published my first cycle map — day by day across a 14-day cycle, built from my own tracking.
  • Daily tracking from 20 August to 15 September: 26 days, morning and evening.
New to download or request
  • The One View as PDF, plain text and JSON — the plain text so a doctor can paste it straight into an email or a note.
  • Daily tracking as a CSV.
  • The 14 September bloods as a PDF, identifiers redacted.
  • The Cycle 3 map as a PDF.
  • The Radical Health report generated from my records.
What it flagged up
  • A genetic specialist in Australia started a conversation about mapping my whole genome rather than relying on panel tests.
  • A team who work on cancer vaccines began looking at whether there's a vaccine approach for my smouldering myeloma.
  • A vaccine specialist in Argentina started sending approaches and research worth taking to my team.
  • Redacted everything before publishing: NHS number, hospital numbers, address, and the names of clinicians and lab staff.