Defense of the AI-Enhanced Mental Health Framework.

The framework is predicated on Human-in-the-Loop governance. The AI functions strictly as a back-end decision-support tool — auditing documentation, tracking follow-ups, and flagging risk markers. It is not the diagnosing agent.

Under the Mental Healthcare Act 2017, clinical liability remains with the licensed professional (the CHO or Psychiatrist). Our protocol mandates a Human Override — no patient is discharged or cleared based solely on an algorithmic output.

The AI provides the visibility; the human retains the accountability.

Our framework is Offline-First. Drawing on the empirical success of the Mukti Program's 4S–6R–5L Protocol, the clinical intervention happens in the village, not in the cloud.

CHOs execute the psychiatric stabilisation and psychosocial support through structured human-led protocols. The digital layer (PANKHUDI) serves as a local repository that syncs whenever connectivity is reached.

We do not require a digital floor to provide clinical care; we require only a human-led protocol — which already exists in our community networks.

We address burnout through Administrative Automation. The current ASHA burden is ~80% manual reporting and ~20% clinical care.

By integrating the AI back-end to automate documentation, reporting, and patient tracking, we reduce the administrative load on ASHAs.

We aren't adding tasks; we are removing the paperwork fatigue that currently prevents them from engaging effectively in community health.

We utilise the Healthy Knots outreach methodology, which treats faith leaders as delivery channels for scientific referral — not as clinical advisors.

Training modules are secular and mhGAP-aligned. We engage multiple faith leaders simultaneously, creating a checks-and-balances ecosystem.

Health outcomes are monitored via our back-end audit layer (SafeMind), ensuring only scientific referrals are facilitated.

Our compliance is Privacy by Design. We have moved beyond long, text-heavy consent forms.

The portal uses multilingual audio-visual consent interfaces so even low-literacy users understand exactly what data is being shared and why.

Architecture stores data locally (data localisation) and provides an automated, ABHA-linked right to erasure — full compliance with the 2023 Act.

The 62% underspend is a result of Administrative Bottlenecks, not a lack of funding. Current DMHP processes are too complex for the overburdened frontline to navigate.

Our framework acts as the absorptive layer — automating the compliance and reporting loop required to unlock these funds.

We turn the DMHP from a reactive, bureaucratic budget into an outcome-based financial engine, proven by the efficiency records of the Mukti Program's community-philanthropy model.