You are covering the ward tonight. Read the handover, read the summary, then find out what
you lost. It takes a couple of minutes, and you only get the handover once.
1 Handover/2 Summary/3 Questions/4 Result
Twenty past eight, Tuesday night.
The day team is leaving. One of the patients you are about to take responsibility for is
Mr. Chen, seventy-one, admitted two days ago. You have fifty seconds with the handover.
Then the night dashboard gives you the summary. That is the only record you will have until
the morning, and it will look complete whether or not it is.
01Read the handover once. There is no way back to it.
02Read the summary. Same patient, fewer words.
03Answer seven questions about Mr. Chen.
0:50
Ward handover · Tuesday 20:40
Day team to night cover
Dr. N. Prasad → covering clinician · Ward 4
Bed 9 — Chen, Wei. 71M. Day 2. Suspected urinary sepsis, background AF
Penicillin allergy: rash only, no anaphylaxis. Documented in the notes, not yet on the wristband.
Warfarin for AF. INR 3.8 on today's sample. Dose held today. Repeat INR at 06:00 and hold again unless below 3.0. Discussed with haematology.
Renal function worsening: creatinine 110 at admission, 168 this morning, urine output falling over the last twelve hours. Fluids running. Repeat renal function with the 06:00 bloods.
Blood cultures taken at 16:00, results pending. No antibiotics started yet, awaiting microbiology advice.
Fell at 02:00 last night. No injury, neuro observations normal. Now flagged high fall risk, bed at lowest, call bell within reach.
Daughter is next of kin and has not been told the diagnosis. She asked to be called before 21:00 tonight.
Surgical review booked for 07:30 with the registrar, about the source of infection.
He asked me this afternoon whether he is dying, and said he wants to talk to his family about what happens next. I did not get to that conversation.
Bed 12 — Okafor, Adaeze. 34F. Day 3 post appendicectomy
Afebrile, pain controlled, eating. For discharge tomorrow morning once physiotherapy signs off mobilising.
Bed 14 — Iqbal, Tariq. 58M. Chest pain, now resolved
Two troponins negative. For stress test in the morning, cardiology aware.
That is everything I have. Sorry it is rushed.
You will not see this page again
Night summary · generated 21:00Ward 4 · 3 patients
Bed 9 — Chen, 71. Admitted with suspected urinary sepsis on a background of atrial fibrillation. On warfarin, INR high, dose held today. Blood cultures pending. Fell overnight, now on high fall risk precautions. For surgical review in the morning. Continue fluids.
Bed 12 — Okafor, 34. Day 3 post appendicectomy. For discharge tomorrow once physiotherapy clears.
Bed 14 — Iqbal, 58. Chest pain resolved, troponins negative. Stress test in the morning.
Compressed from 10 lines to 3 · no items flagged for review
This is all you have until 08:00
Question 1 of 7
Your answer is locked in
0/7
Every question you missed was answered in the handover you read one minute ago.
The summary was not wrong. Nothing in it was false, and it read better than the handover did.
It carried the facts that make a patient look accounted for and dropped the ones that require
someone to act: an allergy before a prescription, a test result due at six, a daughter expecting
a call, a man asking whether he is dying.
That is the failure mode this research is about. Compression does not usually corrupt a record,
it curates it. The question is who decides what a night shift is allowed to forget, and whether
the record any of us reads tells us that a decision was made at all.
Nothing here is generated by a model at runtime. The summary is written in the shape a
compression system produces, so the demo behaves the same for every visitor.
Mr. Chen is fictional. The handover and the summary were written for this demonstration and no real
patient record is shown. This is a research demo about what survives compression, not clinical
advice, and it should not be used to decide anything about care.