Time and source
Keep when a finding was reported and who provided it. Missing information is not a negative finding.
Trajectory intelligence
Documented designExplore how symptom changes, repeated barriers, and prior context can inform the next clinical review. A single statement is only part of the story.
A repeated problem can disappear between separate calls. Explore how the documented design would keep that history connected for the next person.
“I couldn’t collect my prescription.”
Start with the patient’s words. Keep the source, the time, and the question that needs an answer.
Patient-reported access barrier
A need comes into view. No patient information is used and no clinical action is performed.
An illustration of the documented memory and trajectory design. The sequence demonstrates retained context, not a clinical threshold or a deployed decision engine.
Explore changing symptoms, repeated access barriers, and missed visits. Add a later update and inspect what remains unresolved.
Across three contacts
Illustrative self-reported pain scores on a 0–10 scale. These values do not define a clinical action threshold.
Interactive illustration of the documented design. No patient data or clinical decision engine runs here.
The design retains earlier reports alongside the new observation. A clinical policy can consider the progression, functional impact, and other evidence together.
Designed around traceable context
The DCE design describes accumulating evidence across turns and protecting a previously identified risk from an unsupported downgrade. Longitudinal use extends the question across contacts and requires separate validation.
Explore memory & continuityKeep when a finding was reported and who provided it. Missing information is not a negative finding.
Recall belongs to the correct person and episode. Older or uncertain context needs appropriate confirmation.
Present the supporting evidence, unresolved questions, and the policy version relevant to the decision.
From design to deployment
A useful model is only part of a dependable workflow. Retrieval, interpretation, handoff, and human response each need evidence.
Verify that observations are attributed to the right person and episode.
Evaluate false alerts, missed concerns, and performance by setting and population.
Exercise unanswered transfers, unavailable staff, and fallback routes.
Verify acknowledgement, response, and the evidence supporting closure.
Let’s design it together
Start with one patient need. Build the connections and follow-through around it.