M09 · Suite 2 · Continuous Learning Engine
Personalyze™
3 parallel learners · 142 patterns / 7d · 47 active suggestions · 73% acceptance · CAIG-explainable
M09 · Suite 2 · 3-Level AI Learning · Live
✦
Personalyze™
Three parallel learning loops — platform workflow, organization deployment, individual patient.
Teaches Configurator, IRIS, and CVFB. The system gets smarter without ML re-training cycles.
Active Learners
3
L1 + L2 + L3 · all live
Patterns Found · 7d
142
86 actionable
Suggestions Live
47
Configurator · IRIS · CVFB
Acceptance · 30d
73%
▲ +8 vs prior month
The 3 learning levels · running in parallel different inputs · different outputs · different cadences
L1
Platform Workflow
cadence: real-time
Which CVFB workflow templates complete successfully · which routing rules generate high-quality engagements ·
which CLOIE™ triggers fire most predictively.
Outputs CVFB template suggestions
L2
Org Deployment
cadence: nightly
Which Configurator templates produce best outcomes for a given population type ·
which Doppler thresholds match this org's coordinator capacity · cross-org pattern matching.
Outputs Configurator AI suggestions
L3
Patient Behavior
cadence: per-event
Best contact channel per patient · best time of day · language preference learned ·
which IRIS nudge formats this individual responds to.
Outputs IRIS personalization · VIDA timing
Recent pattern discoveries 7 high-confidence · CAIG-explainable · auto-applied to suggestions
L1
Spanish-LEP cohort responds 2.4× better to VIDA vs portal messages
Confidence0.94
Status→ IRIS routing
Applied
L2
MA dual-eligible orgs: lower Doppler thresholds raise gap closure 14%
Confidence0.88
Status→ Configurator
Applied
L3
Eleanor Markham · best contact: VIDA call · 14:00–16:00 · Spanish
Confidence0.96
Status→ VIDA timing
Applied
L1
CHF + COPD comorbidity: weight monitoring beats SpO₂ for 14d-readmit prediction
Confidence0.78
StatusReview pending
Proposed