A practical testbed for earlier risk signalling.

Enter six maternal measurements, run the frozen Random Forest model on the server, and test a simulated caregiver alert under different connectivity conditions.

01 · Input

Maternal measurements

6 variables

AgeyearsModel training support: 10–70 years · plausible: 5–85Systolic blood pressuremmHgModel training support: 70–160 mmHg · plausible: 40–300Diastolic blood pressuremmHgModel training support: 49–100 mmHg · plausible: 20–200Blood sugarmmol/LModel training support: 6–19 mmol/L · plausible: 1–50Body temperature°CModel training support: 36.7–39.4 °C · plausible: 29.4–46.1Heart ratebpmModel training support: 60–90 bpm · plausible: 20–250

Body temperature is entered and reported in °C. The frozen source model was trained with Fahrenheit values, so the server converts Celsius internally at the inference boundary. Values that are plausible but outside training support are not extrapolated — the model abstains and the case is escalated for urgent clinician review.

Simulated connectionalert test

This is a research simulation. Do not enter real patient data or direct identifiers such as names, phone numbers, emails or addresses. Anonymous measurements and results are stored on the server for the prototype audit trail. This is not a clinical patient record.

02 · Output

Assessment result

Ready for an assessment

Complete the six measurements and run the model to see the risk class, probabilities, explanation, and simulated alert status.

Frozen modelServer-side inferencePersisted audit record