Part 1. System context
KPI cards show cohort size and baseline risk. This sets the context: we are dealing with a large chronic-care population, not isolated events.
Overview: What are we presenting in this dashboard?
We are not analyzing a single diagnosis only. We are presenting how diabetes patients move through the healthcare system over time, how social and geographic factors shape that journey, and how this creates unequal system burden.
Part 2. Fragmentation and early warning
Model and capture charts show how well we flag future ED events or long gaps. If top-risk capture is high, care teams can triage limited outreach more effectively.
Part 3. Structural inequality patterns
Cluster charts identify pathway archetypes: long-gap continuity failure, stable follow-up, and high-ED pressure groups. This shows inequality is structural, not only individual outliers.
Part 4. Action and policy direction
Geographic hotspots and patient drill-down turn results into action: continuity interventions for long-gap groups, and access interventions for high-ED groups.
How reliably can we flag upcoming care breakdown risk?
For each extra 5% outreach capacity, how many future events do we gain?
Bars show incremental yield by bucket (0-5%, 5-10%, ...), and the dashed line is random selection benchmark. Higher-than-random bars mean better triage efficiency.
Which clusters show continuity breakdown and emergency pressure together?
Bars show mean inter-visit gap; line shows ED ratio. This reveals whether fragmentation and acute burden co-occur in the same journey group.
Where does cluster-level burden concentrate?
Each bubble is a cluster: farther right means longer gaps, higher means higher ED ratio, and larger means more patients.
Which neighborhoods have the biggest risk gap?
Bars compare predicted vs observed risk rate; the line shows prediction gap (percentage points). Larger absolute gap means higher calibration uncertainty in that neighborhood.
Which individual patient story should we review next?
Open one patient timeline for case-level review. You can type an ID or click a patient ID in the watchlist table.
Who should we contact first under limited outreach capacity?
| Patient | Encounter | Date | Risk Score | Observed Risk | ED ≤ 90d | Gap > 120d |
|---|