Every cancer that stops responding to treatment escapes through a door. Some cancers hide the target the drug was aiming at. Some mutate the switch the drug was blocking. Some simply outgrow a dose that was too small. Oncologists know these doors well — the hard part is that a tumour usually has several doors open at once, and no single drug covers them all.
What the simulation actually does
When we analyse a patient’s records, we start with their tumour’s own test results: which genes are broken, which targets are present, how fit the patient is, what their bone marrow and liver can tolerate.
Then the engine does something a human cannot do at a whiteboard: it takes a library of real, available drugs and scores every allowed combination — in recent cases, between 169 and 440 of them — against the specific escape routes that patient’s cancer can use.
Each combination gets a score for:
- Coverage — how many escape doors it closes at the same time
- Evidence — how strong the clinical trial data is for each drug in this cancer
- Safety — combinations that would overload the patient’s liver or marrow are penalised or excluded outright
What comes out the other end
A ranked list — and, more importantly, the reasoning behind it. In one lymphoma analysis, every top-ranked plan was built around the same backbone, because only that class of treatment covered the escape route that sinks everything else. The striking part: the patient’s own doctor had reached the same conclusion from the same evidence. The machine and the physician agreed on what — the difference was only about when.
An honest map, not a prescription
We say this on every report and we will say it here: the ranked list is a map, not a prescription. Some highly-ranked options are investigational, or approved only in other cancers. What the map is good for is the conversation it creates — it lets a family walk into an oncologist’s office asking sharper questions: which escape routes does my current plan cover, and which are still open?
That question, asked early, is worth a great deal.