Who is available on Saturday? One answer the whole club shares.
Drop the scan report in; a coded case comes out. Coach and physio read the same case — so they can never disagree.
Two versions of the truth
Coach has a spreadsheet. Physio has a notebook. By Thursday they disagree.
Under pressure, someone gives a date nobody can defend.
He returns on the calendar, not on criteria. He re-tears.
The board is the caseload
Derived from the open cases, so it cannot disagree with the physio.
| Player | Case | Window | Status |
|---|---|---|---|
| A. Whitfield | Hamstring — biceps femoris, grade 2 | 18–24 d | Out |
| J. Okoro | Lateral ankle sprain — grade 1 | 6–11 d | Doubtful |
| M. Sandoval | Groin — adductor longus | criteria met | Available |
| T. Beaumont | Concussion — day 4 | no day count | Out |
| R. Iversen | Calf — awaiting grading | — | Unconfirmed |
Nobody has checked in 48 hours — shown grey, not a colour it can’t vouch for.
No day count, ever. Graduated, symptom-limited return.
A clinician’s call against exit tests.
The spike, three weeks early
Acute:chronic workload on Gabbett’s model, from session RPE × minutes.
Computed from the cases — not typed into a second list.
The number never travels alone
Every figure carries its tier, its citations and a full lineage — contributing records, and the ones set aside, with the reason.
- Condition
- Hamstring, biceps femoris — grade 2 · TM1
- Cohort
- Adult · male · elite · football — matched on sport, sex and age band
- Endpoint
- Time to return to play — not full training, not symptom resolution
- Contributing
- 2 records pooled by sample size · 1 weaker-tier kept as context, never averaged in
- Set aside
- 1 record — different endpoint, with the reason recorded
It knows when not to answer
A plausible number is more dangerous than none. Someone returns to play on it — so refusal is a designed output.
No grade recorded. Returns the options, issues no figure.
No evidence for this sport, sex or age band. Says so rather than borrowing.
A revision is not a primary repair.
Evidence altered → everything fails closed.
180 records. The strongest tier governs.
Pooled by sample size where studies agree; weaker studies become context, never averaged into the number.
A symptom-resolution number and a return-to-match number are months apart.
Five gates before a clinician sees it
Europe PMC, with the exact query shown.
Same question, same population, permanent ID.
Two independent passes, compared field by field.
Checked against source, committed by name.
Corpus signed. Checked at serve time.
59 of 161 conditions. The rest refuse.
The other 102 are known to the engine and refuse rather than estimate. That ratio is the roadmap.
“Governed” is a claim. This is the check.
For every discharged case: what the engine said (frozen at case opening), what the clinician decided, and what actually happened.
Read from the day the clinician saw it — never recomputed. A ledger that regrades history marks its own homework.
Time to discharge, not first match back — so the caveat travels with the number.
No day count is issued, so there is no prediction to grade. Excluded, in the ledger.
Below 5 cases the aggregate is withheld. Medians, not means.
Board, load series, figure and body map are illustrative; player names are not real. Evidence tiers, the five return-to-play clocks, coverage and code counts are measured against the running system. Decision support for the treating team — a clinician commits every clinical decision.
Get Injury IQ switched on
We are taking a small number of pilots so the product is shaped by people using it rather than by us guessing. Products are enabled per organisation, so one request covers your whole club.
A request, not an account — we enable products per organisation, so a person tells us who they are and we come back to them.