CURRENT EVENT
Waiting
LIVE SANDBOX
Run the operating-room day, not just the paper.
Change the emergency mix, workload, and resource disruptions. Then schedule patients yourself or let a policy make the next move. The interface exposes the trade-off between throughput, emergency access, utilisation, and waiting-time equity.
Educational replay
This sandbox is calibrated to values reported in the article. It is not a re-execution of the authors’ source code and should not be interpreted as clinical decision support.
DAY–ROOM LATTICE
Operating rooms
electiveurgentemergency
LIVE RECOMMENDATION
Waiting for selection
Select a patient to see guidance
The application will compare all currently feasible OR slots using an educational scoring layer inspired by the selected policy, current capacity, urgency, waiting time and disruptions. The optional specialised-room mode is exploratory and was not evaluated in the article.
Recommended
Acceptable
Higher trade-off
Not feasible
NARRATOR / AUDIT LAYER
What happened after the assignment?
Select a patient. In Guided mode, the open slots will be ranked before you place the case.
COORDINATION FIELD LAB
See how the hospital “signals” where a patient should go.
Each cell represents one operating-room/day combination. The shared field shows illustrative normalised signals for demand, urgency, free capacity, conflict pressure, and aggregate waiting-time equity pressure.
Stigmergy, in one sentence
Agents do not need to message each other directly: they alter a shared environment, and the next decision reads those signals.
1DepositQueue pressure is written into the shared field.
↓
2DiffusionPart of each signal spreads to neighbouring day–room cells.
↓
3DecayOlder signals gradually lose influence.
↓
4RecommendationClick a cell or deposit the queue to reveal the strongest candidate.
DAY × OR HEATMAP
Urgency pressure
LowHigh
OR 1OR 2OR 3OR 4OR 5
FIELD DYNAMICS
Change how signals spread
Higher decay removes old signals faster. Higher diffusion spreads local pressure more strongly to neighbouring cells. The app automatically limits ρ to the article’s sufficient stability bound, (1−λ)/4, for a four-neighbour lattice.
Show model equation
Φdr(n+1) = (1−λ)Φdr(n) + ρΣ(d′,r′)∈Ndr(Φd′r′(n)−Φdr(n)) + Zdr(n)
Current controls: λ = 0.30 · ρ = 0.15 · threshold τ = 0.01. Exact values are shown for transparency; the heatmap displays LOW / MEDIUM / HIGH.
PAIRED SCENARIO
Put two policies on the same patient stream.
Use the same emergency mix and workload for both policies. This mirrors the article’s paired-comparison logic and makes trade-offs visible rather than hiding them behind one score.
Same scenario, different ruleReported article values anchor the baseline. Intermediate slider positions are illustrative interpolations.
VS
EVIDENCE LENS
The result changes when the scenario changes.
The application keeps the article’s most important limitation in view: the synthetic 15.2% emergency mix and the MIMIC-III Demo-derived 78.4% mix do not tell the same story.
NeuroCare-H scheduling rate
In the main synthetic scenario, NeuroCare-H scheduled 92.47% of cases, compared with 87.68% for FCFS.
Emergency waiting
Emergency wait was 0.01 days under NeuroCare-H in the synthetic evaluation, versus 0.40 days for FCFS.
Gini trade-off
Waiting-time inequality was higher with NeuroCare-H than FCFS (0.63 vs 0.36), so efficiency gains did not imply better aggregate equity.
MIMIC-derived reversal
With the 78.4% emergency mix derived from MIMIC-III Demo, the NeuroCare-H versus FCFS scheduling-rate difference was no longer statistically significant.