Skip to main content

ReAct Loop Architecture

Chorus incorporates a proactive, self-adjusting reasoning workflow model inspired by the Reason–Act–Observe (ReAct) paradigm.

Continuous Situational Awareness

This iterative loop enables the system to adapt to changing patient conditions dynamically and maintain continuous situational awareness across large monitored populations, without overwhelming clinicians.

The ReAct Cycle​

1. Reason​

Incoming physiological data is analyzed continuously against clinical baselines. The system uses machine learning models and deterministic rules to determine whether patterns indicate normal monitoring status or potential clinical concern (e.g., compounding weight gain and rising blood pressure for a heart failure patient).

2. Act​

When defined thresholds or model predictions are met, the system initiates an automated workflow action.

  • High Risk: Generating alerts or requesting immediate clinician review.
  • Medium Risk: Updating patient risk stratifications in the registry.
  • Low Risk: Automatically logging a billing event or silently continuing.

3. Observe​

Subsequent patient data is evaluated to determine whether the intervention (e.g., a medication change made by the doctor) or natural observation changes the patient's risk profile. This feeds directly back into the Reason step.