Hospital orchestration command center

The hospital, coordinated — not just displayed.

Pulse fuses patient flow, beds, staffing, and care coordination into one live operating picture, then runs a transparent optimization brain that turns the current state of the building into ranked, justified, one-tap actions — placing a boarder, backfilling a sick call, expediting a discharge. It alerts caregivers, connects to hospital systems, and continuously optimizes operations to remove the administrative overhead a hospital runs on today.

~186 beds, one source of truth Bottlenecks predicted hours ahead 27 modules across 12 domains
The problem

The invisible coordination overhead

A hospital is a real-time logistics system pretending to be a set of departments. The work of coordinating it — who goes in which bed, who covers the sick call, what's blocking this discharge, which waiting patient is quietly deteriorating — is done today by phones, pagers, whiteboards, spreadsheets, and a morning huddle that's stale by mid-morning. That overhead is enormous, invisible, and exactly what software should absorb.

Phone tag and pager relays
Every placement, consult, and coverage gap chased one call at a time, with no owner, no clock, and no record.
Whiteboards and the stale 7am huddle
A single shared picture of the house exists only on a wall and only first thing in the morning — wrong by lunch.
ED boarding
Admitted patients hold in the ED for hours waiting on a clean, appropriate bed — the costliest, most preventable backup.
Left without being seen (LWBS)
Patients walk out of an overwhelmed waiting room — lost care, lost revenue, and a safety risk no one sees in real time.
The sick-call scramble
A 5am call-out triggers a frantic, ad-hoc hunt — and agency staff get called before cheaper, better-fit options are even tried.
Manual bed placement
Matching a patient to a bed means juggling isolation, telemetry, gender, and service rules in someone's head, under pressure.
Who it's for

Built for the people who run the building

The primary user is the hospital operations command center — the charge nurses, bed managers, and staffing coordinators who keep the house moving. Everyone downstream benefits, and so do patients and families.

Primary user

Ops / Command Center

One role-aware operating picture: census forecast, surge index, the ranked action queue, and live KPIs — the place where the whole house is coordinated from.

Primary user

Charge Nurses

A throughput board with live wait timers, door-to-provider forecasts, a deterioration watchlist, and one-tap rapid-response escalation — alerts that actually matter.

Primary user

Bed Managers

Constraint-aware ranked placement with one-tap assign on a reconciled real-time board — accepting a placement fans out EVS, transport, and the receiving-unit alert automatically.

Primary user

Staffing Coordinators

Auto-detected coverage gaps run a cost-and-fit-ranked, fatigue-aware open-shift cascade — same-unit, float, OT, agency last — with one-tap fill and live ratio gauges.

Downstream owner

EVS & Transport

Demand-prioritized cleaning and transport jobs with smart dispatch, ETAs, and stall escalation — the bed-turn loop closed instead of paged.

Executive view

Executives

The closed-loop impact ledger — boarding hours avoided, moves and agency shifts saved — plus surge readiness and the KPIs that tie operations to outcomes and cost.

Patients & families benefit indirectly

When the building runs smoothly, waits get shorter and more transparent, deterioration is caught earlier, placement is faster, and discharge is smoother. Pulse never touches the patient directly — it removes the friction that was slowing their care.

What it does

Five flagship modules, one operating picture

Each screen ends in an action, not a number to stare at. These five are built and clickable in the prototype; they share one store, one brain, and one urgency language.

Mission Control

The single operating picture: census & surge gauge, the brain's ranked action queue, the anomaly Watchtower, the closed-loop impact ledger, and the live command feed.

ER & Patient Flow

Per-patient and per-acuity predictive wait and door-to-provider forecasts (p50/p90), a deterioration watchlist with threshold auto-escalation, and an ED-boarding forecast.

Beds & EVS Turnover

Constraint-aware ranked bed placement with one-tap assign, a live 186-bed board, and an EVS smart-dispatch Kanban. Accepting fans out EVS, transport, and the receiving-unit alert.

Staffing Sick-Call Backfill

Auto-detected coverage gaps run a cost-and-fit-ranked, fatigue-aware open-shift cascade (same-unit → float → OT → agency) with one-tap fill, plus live acuity-weighted ratio gauges.

Command Feed

Every alert and recommendation lifecycle event as one closed-loop stream — owner routing, SLA timers, acknowledgment, and auto-escalation ladders.

+ 22 more in the catalog

Discharge & barriers, perioperative, diagnostics/TAT, safety & quality, transport & assets, revenue & prior-auth and more — 27 modules across 12 domains, all wired to the same stores.

The five flagships are part of a 27-module catalog across 12 domains — patient flow, beds, workforce, clinical care, periop, diagnostics, supply & assets, patient experience, safety & quality, command-center ML, integrations, and revenue cycle.
How it works

The closed orchestration loop

The measure of success isn't dashboards — it's administrative overhead removed and throughput recovered. Pulse runs one loop, continuously, and proves it.

Ingest live signals

Arrivals, admits, discharges, sick calls, deterioration, EVS completion — the signals that already exist, as one normalized stream.

Predict bottlenecks

Explainable predictors forecast where backups form hours ahead — census, boarding, wait quantiles, deterioration, surge.

Propose ranked actions

Solvers turn the forecast into specific, ranked, fully justified recommendations — every one carries its reasoning.

Route to the owner

Each action lands with the right person, an SLA timer, and an escalation ladder — no phone tag, no orphaned task.

Fan out on accept

One tap (human-gated) executes and cascades the downstream work: a placement spawns EVS, transport, and a receiving alert.

Track outcomes

Every recommendation is tracked proposed → accepted/overridden → executed → outcome, tallying hours and shifts saved.

The brain is transparent

It is not a black box. Each predictor is an explainable weighted sum of normalized features, and every recommendation ships its score breakdown so the UI can render exactly why it ranked where it did.

isolation match × 0.30 = +0.27
telemetry fit    × 0.25 = +0.25
minimize moves × 0.20 = +0.18
ratio balance   × 0.15 = +0.12
— — — — — — — — — — — —
score = 0.92

It never auto-executes

Every action is always visible and always overridable. The brain ranks and justifies; a human decides. The one-tap accept is fast, but the authority — and the override — stays with the caregiver.

Human-in-the-loop, by design

One urgency language — ok · watch · urgent · critical — means a dirty-too-long bed, an SLA-at-risk wait, and a ratio breach all read identically across every board.

Benefits

Value for the whole hospital

The same loop pays off three ways at once — for the institution, for the people who work in it, and for the patients and families it serves.

Hospitals

Throughput, capacity, cost, readiness
  • Recovered throughput and faster patient flow
  • Higher safe occupancy without compromising care
  • Fewer avoidable patient-days and shorter LOS
  • Less agency and overtime spend
  • Surge readiness — see and absorb spikes early

Staff

Less friction, fairer load, real alerts
  • Far less phone tag and ad-hoc coordination
  • Fatigue-aware and fair staffing decisions
  • Faster placement with constraints handled for you
  • Alerts that actually matter, with owners and SLAs
  • One shared picture instead of six whiteboards

Patients & families

Shorter waits, earlier safety, smoother care
  • Shorter and more transparent waits
  • Leave-and-return instead of sitting in a waiting room
  • Earlier detection of deterioration
  • Smoother, less-delayed discharge
  • A calmer, better-coordinated experience overall
How it makes the hospital more efficient

The KPIs it moves

Pulse steers the operational metrics that compound into capacity, safety, and margin. These are representative of the dial it turns.

ED boarding hours▼ down
−31%
Door-to-provider▼ down
−18min
LOS / avoidable days▼ down
−0.4days
Bed-turn time▼ down
−27%
Discharge-by-noon▲ up
+22pts
Ratio compliance▲ up
+14pts
Agency / OT spend▼ down
−19%
30-day readmissions▼ down
−9%
Figures are illustrative directional targets for a prototype, not measured results. Real impact must be proven against a baseline, not secular trend.
Responsible by design

Trustworthy where it counts — and honest about the rest

For a system that allocates scarce beds, shifts, and discharges, provable fairness and human authority aren't a footnote. A faithful concept also names what it does not yet handle.

Human-in-the-loop
Nothing auto-executes silently. The brain proposes and justifies; a person accepts or overrides, every time.
Explainable recommendations
Every action ships its reasoning as factor × weight = contribution — auditable, not opaque.
Fairness & override governance
Subgroup calibration, override-rate as a model-health signal, and explicit autonomy boundaries.
Integration-shaped from day one
The store contract mirrors a real ADT/HL7/FHIR feed, so simulation can be swapped for live data without touching the UI.

Known gaps a real build must own

Named honestly — these are not yet handled by the model.

  • Service lines missing — pediatrics, NICU, and L&D/obstetrics (couplet care, non-deferrable births) need their own constraints and forecasts.
  • Model bias & fairness — triage, deterioration, no-show, and denial-risk models carry documented demographic-bias risk (e.g. pulse-ox by skin tone) that needs first-class audit.
  • Consent & surveillance ethics — RTLS, hand-hygiene sensors, fatigue scoring, and family geofencing are pervasive monitoring that need consent, data minimization, and labor agreement.
  • Downtime resilience — a command center that goes blind when feeds fail or ransomware hits is itself a safety risk; it needs a graceful-degradation playbook.
  • ROI attribution & TCO — avoided boarding hours must be proven against a baseline; every EHR interface is a real integration project, not a switch.
Roadmap · beyond the catalog

The differentiated ideas

The defensible bets that go past the obvious operational lens. None are built yet, but the architecture is designed so each slots into the same event bus and brain.

Flow

Behavioral-health & psych-boarding lane

A distinct flow for the worst real boarding pathology — external psych-bed search, involuntary-hold legal clocks, elopement risk, sitter optimization, and ligature-safe placement constraints.

Resilience

Degraded-mode / downtime command

A first-class business-continuity posture: the wall switches to a downtime view, drives paper-fallback reconciliation, and orchestrates manual catch-up on recovery.

Network

Regional transfer center

Mission-control of mission-controls — system-wide bed availability, EMS destination steering, outside-hospital transfer triage, and repatriation across nodes.

Outcomes

Readmission prevention

Turn the 30-day readmission KPI into an owned workflow — risk-driven interventions and PCP/SNF/home-health handoff for the most expensive failure mode.

Emergency

HICS mass-casualty mode

A distinct command posture from ordinary surge — START triage, decon/lockdown/evacuation orchestration, external-event feeds — satisfying CMS Emergency Preparedness.

Governance

ML fairness governance

Subgroup calibration dashboards, override-rate as model health, and explicit autonomy boundaries — provable non-discrimination as an ethical necessity and a procurement edge.