Live ML Prototype

DigitalGuardian Core

Data Stream Input

Adjust physiological parameters to simulate student state.

Sleep Quality50%
Heart Rate Var (HRV)50ms
Physical ActivityModerate
SYSTEM READY. WAITING FOR INPUT...
HRV (STRESS RESILIENCE)
SLEEP EFFICIENCY
High Risk
Stable
?
⚠️ High Risk Detected Classification Result The AI mapped this student into the cluster of previous high-stress patterns.

Future Expansion: From Stress to Suicide Prediction

This prototype is trained on the SWELL Dataset (Stress Detection) with 92% Accuracy. To upgrade this to a clinical-grade Suicide Prediction System (99% Accuracy), we need funding to access the following restricted medical datasets:

1. UK Biobank Study (Gold Standard)

  • 15,768+ participants with ECG/HRV data
  • Real suicide attempts tracked over years
  • Barrier: £6,000+ Fee & 6-month Approval

2. Adolescent Inpatient Studies

  • 51 acutely suicidal teenagers (Ages 12-18)
  • HRV measured directly via Fitbit
  • Barrier: Protected Clinical Data (Requires Funding)

3. Wearable IoT Research

  • Proven 85-92% accuracy in pilot studies
  • Combines Mood Ratings + Physiological Data
  • Barrier: Not Publicly Downloadable
"We have the algorithm (SWELL). investment gives us the Data (Biobank) to save lives.This data turns a 'Concept' into a 'Life-Saving Medical Device"