This project is testing the efficacy of a 12-week remote therapy intervention (RTI) to decrease aggression and victimization behaviors among violently injured youth (age 14-20) seeking treatment in the emergency department (ED) in Detroit or Flint.

Study Design

The intervention has 3 study arms:

  1. S-RTI: Standard ED counseling session + 5 post-ED remote therapy sessions
  2. AI-RTI: An adaptive artificial intelligence (AI) supported RTI package that optimizes therapeutic intensity using reinforcement learning. Participants are assigned to varying therapy conditions based on their risk status each week. Therapies in this arm include:
    • 30-minute telephone-based remote therapy session;
    • Automated electronic tailored therapy (chatbot); or,
    • Assessment only without intervention (i.e., none)
  3. Standard care

Randomization is stratified by age (14-17; 18-20) and sex.


CHCR leveraged our MTS system to create highly tailored, interactive chatbot interactions in the AI-RTI arm for automated electronic tailored therapy. The chatbot was also designed to show different emotions to help build empathy and rapport with participants. Participants saw relevant testimonials, infographics and behavioral tips within the chatbot as well. The chatbot was delivered on a personalized website that also included links to community resources.

Therapist MI Dashboard

CHCR also created a therapist dashboard to guide the ED counseling session based on principles of motivational interviewing (MI). Data from a participant surveys and chatbot responses was fed into this feature in order to produce tailored summary screens to help guide the counselor’s MI session with that participant. The therapist dashboard also records all relevant data for each counseling session.