VigilantAir — AI-Assisted Anaesthesia Patient Monitoring
Streams multiparameter vitals into ML models that forecast complications early and issue context-aware alerts with reduced false alarms, supporting timely intervention in theatre.
Overview
VigilantAir is an AI-driven decision-support layer for the anaesthetist. It ingests live streams of HR, SpO₂, BP, RR, EtCO₂ and other parameters from standard OT monitors, cleans and aggregates the signals, and then applies time-series models to recognize trajectories associated with clinical risk. Instead of flooding the team with alarms, VigilantAir focuses on early, context-aware warnings that explain why attention is needed and visualizes the recent trend so that action can be taken before decompensation.
Clinical Problem
In long or complex cases, important trends can be subtle and build over time. Alarm fatigue from threshold-only triggers reduces trust, sometimes causing teams to dismiss truly significant alerts. There is a need for an intelligent layer that separates physiological variation from clinically meaningful change, forecasting risk with enough lead time to intervene, and doing so without adding cognitive overload.
Methodology
- Adapt connectors to ingest live data from standard multiparameter monitors; clean signals and engineer features for stability.
- Train and tune time-series ML models using guideline thresholds and outcome-tagged examples.
- Present on-screen, audio and mobile notifications with a brief rationale and trend visualization to support timely action.
Tech Stack
Equipment
Expected Outcomes
Early model prototyping; controlled OT pilot planned after internal validation.