Description: Paste the following summary (I drafted this based on your paper's abstract and introduction): This research presents a platform-neutral, domain-specific generative AI framework engineered exclusively for structured mental health triage among Indian adolescents (ages 16-18). Addressing the massive treatment gap and socio-cultural stigmas in India, the system adopts a novel "disengagement-oriented" design. It utilizes strict architectural constraints—including deterministic interaction limits, real-time crisis detection, and mandatory non-therapeutic conversational boundaries—to mitigate the ethical risks of Large Language Models (LLMs), such as deceptive empathy and parasocial attachment. The AI functions strictly as a transitional gateway, classifying distress severity across culturally relevant domains (e.g., "overthinking," "academic stress") and enforcing automated referral pathways to licensed human professionals or national tele-health infrastructures like Tele-MANAS.
Manav Shah (Wed,) studied this question.