BedFlow Command AI

• 

  Human Governance: authorised staff review, approve, or reject recommendations, creating an accountable audit record.

• 

  AI Advisory Context: provides policy-aware operational guidance to support staff review; it does not make clinical discharge decisions.

• 

  BedFlow Command Center: a simulated hospital-operations view of capacity, expected discharges, ED boarding, and pending operational work.

• 

  Patient Case Review: shows a simulated case with discharge destination, unresolved blockers, readiness checklist, ownership, and task status

• 

  Explainable Action Plan: turns operational signals into a plain-language explanation, estimated delay impact, and structured next steps.

• 

  Prioritised Queue: ranks non-clinical discharge blockers—including insurance, transport, placement, and medication reconciliation

• 

  What-If Baseline: a sandbox showing the baseline projected delay, ED boarding, expedited-review candidates, and workflow beds recovered.

• 

  Saved What-If Scenario: tests operational changes—such as insurance, transport, EVS, two case managers, and five surge beds

Inspiration

Hospitals need to move patients safely while balancing discharge readiness, readmission risk, and limited bed capacity. BedFlow Command AI helps make those decisions clearer without removing clinician control.

What it does

BedFlow Command AI reviews hospital-flow signals and highlights patients who may be ready for discharge, need extra review, or should be prioritised for bed-flow action. Every recommendation is explainable and requires human approval.

How we built it

We built a new 2026 BedFlow dashboard using Python, Streamlit, predictive risk/readiness logic, and an AI-assisted decision workflow. The interface turns patient-flow data into clear recommendations for clinical teams.

Challenges we ran into

The main challenge was ensuring recommendations were useful but not treated as automatic clinical decisions. We designed the workflow so AI supports staff with evidence and explanations, while authorised people remain responsible for final action.

Accomplishments that we're proud of

• Created a focused hospital-flow command dashboard
• Made patient-priority recommendations understandable
• Put human approval directly into the decision workflow
• Demonstrated responsible AI for a real operational healthcare problem

What we learned

A strong healthcare AI product needs transparency, accountability, and a workflow people can trust—not just accurate predictions.

What's next for BedFlow Command AI

We plan to add scenario planning, audit history, role-based approvals, live data integrations, and more evaluation with hospital operations users.

Built With

• explianable-ai
• generative-ai
• healthcare-ai
• human-in-the-loop
• pandas
• python
• scikit-learn
• streamlit
• xgboost