AI Studio
Predict
Mock AI models and datasets that will be used for risk forecasting and alert intelligence. No real patient data is processed in this MVP.
“High modern” preview
AI models
3
Forecasting, classification, detection
Datasets connected
4
Wearables, IMU, compliance, EHR
Predictions ready
1 pipeline
Risk forecast (7-day horizon)
Risk forecast (next 7 days)
Baseline
Predicted
Feature signals (mock)
What the model “pays attention to” for the current forecast.
SpO₂ variability
Increased variance contributes to higher risk.
Resting heart rate
Elevated baseline over last 72h.
Wear-time compliance
More gaps increase uncertainty and risk.
Sleep regularity
Stable sleep schedule reduces risk.
AI models
Deployed services powering predictions.
CarePulse Risk Forecaster
Risk forecasting • v0.9.2
AUC
—
F1
—
MAE
6.8
Alert Classifier
Alert classification • v0.7.4
AUC
0.93
F1
0.86
MAE
—
Vitals Anomaly Detector
Anomaly detection • v0.6.1
AUC
0.9
F1
—
MAE
—
Datasets
Sources and schemas used by the AI pipeline.
Vitals • 1-min aggregates
Wearable • Updated Feb 10, 08:25 AM
18M rowstimestampheartRatespo2temp+2 more
Falls • IMU events
IMU • Updated Feb 10, 08:11 AM
64K rowstimestampaccelgyroimpact+1 more
Device compliance • wear time
Device telemetry • Updated Feb 10, 08:19 AM
3M rowstimestampbatterywornsignal+1 more
EHR snapshot • meds & conditions
EHR • Updated Feb 09, 11:40 PM
13K rowspatientIdconditionsmedicationsrecentVisits