Sapio HealthSapio Health
  • Today
  • Patients
  • Inbox
  • Calendar
Insight
  • Population
  • Quality (QOF)
System
  • Models
  • Data feeds
  • Settings
Practice trajectory
-26%
JD
Dr John Dolittle
Ashford Vale Medical Group
All engines
Service 3

Horizon-1

Continuous physiological simulation
Version
2026.04.2
Retrained
2026-03-29
Status
healthy
Performance

Key metrics

Calibration & discrimination against the engine's published target. Under-target values are highlighted.

Calibration (Brier)
0.089target 0.100
1y AUC (decompensation)
0.862target 0.850
5y AUC (MACE)
0.814target 0.800
Coverage (axes)
92.0%target 90%
14-day inference volume
1,386 in the last 24h · avg latency 214 ms
Metrics are tracked on held-out evaluation cohorts; values here reflect the last retrain.
Model

Training basis & runtime

Transparent about what the engine was trained on and how it behaves in production.

Model family
Mixture-of-axes neural ODE ensemble
Trained on
115,489 patients · 412M datapoints
Active trajectories
2,841
Mean confidence
27 / 40
across the practice
Avg rebuild latency
214 ms
Projection horizon
10y
History

Changelog

Each retrain is documented; clinicians can see what changed and when.

  1. 2026.04.22026-03-29
    Added ApoB and Lp(a) into the lipid sub-manifold; retrained on expanded cohort.
  2. 2026.01.02026-01-11
    Introduced MASLD sub-manifold; FIB-4 and liver-fat surrogate axes.
  3. 2025.10.12025-10-08
    Calibration improved on elderly cohort after adding frailty adjuster.
Safety

Guardrails

Hard limits and runtime checks the engine enforces.

  • Never emits a projection past 10 years.
  • Blocks rendering when observational confidence <15/40.
  • Uncertainty bands widen automatically as data sparsifies.
Observability

Recent incidents & retrains

Automatic monitoring — warn / critical thresholds on calibration, drift, and latency.

  1. Nightly retrain completedinfoHorizon-116/04/2026, 04:12
    Horizon-1 2026.04.2 retrain completed in 38 minutes · 412M datapoints · calibration within tolerance.