RapidMedX — Smart Patient Triage
AI-assisted triage for hospitals that reduced wait times and improved outcomes.
Problem
Large hospitals struggled with long ED wait times and inconsistent triage prioritisation. Clinicians needed an accurate, fast tool to rank incoming patients by risk so critical cases were seen earlier.
Our Approach
We designed a lightweight clinical AI that augments triage nurses by analysing presenting complaints, vitals, and historical records to produce a real-time risk score. The system integrates with the hospital's EHR and provides clear explanations for each risk signal.
Solution
- Built a diagnosis-aware NLP model to parse triage notes and symptoms.
- Fused symptom embeddings with vitals and prior history using a hybrid ML model.
- Deployed as an edge-capable microservice with EHR connectors and a nurse-facing dashboard.
Outcome
In pilot deployments across three regional hospitals:
- Average patient wait time reduced by 40%.
- Prioritisation accuracy increased by 18% (clinician-validated).
- Adoption rate among triage staff > 85% within 6 weeks.
Tech Stack
Python, FastAPI, TensorFlow/PyTorch (model ensemble), PostgreSQL, Docker, MQTT for device telemetry, React dashboard.
Client Impact
Improved throughput, better clinical outcomes for high-risk patients, and measurable staff time savings.