SIGNAL
Shifting patient behavior from self-diagnosis to structured observation: a B2B2C design exploration in healthcare behavioral design.
Signal is a behavioral intervention designed as a product: It aims to lower avoidable Emergency Department (ED) costs for insurers and mitigate consultation inefficiency for clinics by shifting patient behavior from self-diagnosis to structured observation in the hours and days before they seek care.
Entry: The Circuit Breaker
Decision: Three-phase breathing before every recording.
Stress and anxiety distort how patients perceive and describe symptoms. This deliberate friction "breaks" the sympathetic activation, leading to more consistent and objective descriptions.
Speak freely. The AI structures it. Patient confirm.
Decision: Unstructured voice capture
Voice-First Literacy
Decision: Showing the AI’s "thinking" steps
Writing under stress requires rational processing that anxiety often suppresses; speaking has lower friction. The app builds clinical vocabulary as a side effect of use.
The Algorithm: Direction over Diagnosis
Decision: A two-layer pattern recognition system providing three directional outputs: Monitor, Advise, or Act.
The algorithm does not apply clinical protocols or name conditions. Instead, it functions as a structured support tool that identifies escalation
In an emergency, every additional interaction is a barrier. The interface pivots to a single high-contrast Call-to-Action (CTA) for emergency services and an ephemeral link for medical staff, streamlining the transition to professional care.
Visualization: Deltas over Data
Decision: Highlighting only what changed between observations.
A single symptom is just a data point; a sequence is a picture. By striking through old values and highlighting new ones, the interface trains the patient to read their own patterns of evolution without needing clinical terms.
