🚨 Predictive Crisis Model (Early Warning)

עקרון: ML model identifies "who's about to drop out?" weeks in advance. Early intervention before crisis.

Crisis Risk Score

23%

Risk Level

Medium

Timeline

Next 2-3 weeks

Contributing Signals

Missed 3 sessions last week

+8%

EMA mood declining (6→4)

+7%

Sleep pattern disrupted

+5%

Peer in cohort quit

+3%

✅ Recommended Interventions (NOW):

• Call parent: check for stressors

• Gamify next session (increase reward)

• Pair with motivated peer (accountability)

💡 Impact:

Early detection → prevent dropout. 1 proactive phone call = $5K+ saved in re-intake costs.

base44
Edit with Base44