The 10 Year Health Plan, read as a problem set
I am an NHS clinician. This section starts from the service, not from the hardware. The Plan commits England to three shifts; each one implies a question about allocation, prediction or simulation. Some of those questions are genuinely combinatorial. Most are not, and saying so is the point of the page.
Where a question is combinatorial, it is stated formally, encoded, and run on the Selene emulator against a classical baseline. Every result on this site is an emulator result on synthetic data, and on every problem so far the classical solver wins.
The three shifts
A single patient record, an NHS App that actually does something, and AI in the clinical workflow.
Read →Predict and prevent, with genomics and population health doing the predicting.
Read →Neighbourhood health services, care as locally as possible — and the bed that cannot be freed.
Read →Two settings
The shifts land in two places a patient recognises: the surgery and the hospital. Each page separates the problems a solver could touch from the ones that are really about capacity, workforce and social care.
The flagship problem
One problem gets the full treatment: delayed discharge. It sits squarely in the third shift, it is measurably the largest source of lost bed capacity, and it is the one place where the structure is honestly a constrained assignment problem rather than a staffing shortfall dressed up as one.
A 12-variable discharge-placement instance with an equity term, solved exactly, greedily, by annealing, and by QAOA on Selene. Gate G22.
Read →20 July – 19 November 2026 · the five R&D themes mapped onto NHS problems and onto the gates already built here.
Read →Sources
- Fit for the Future: 10 Year Health Plan for England (DHSC, July 2025)
- BMA — NHS hospital beds data analysisbed stock and occupancy
- NHS England — Urgent and Emergency Care daily situation reportsno-criteria-to-reside series, updated weekly
- NHS England — Core20PLUS5 (adults)