NIMHANS-led HEADS Project to Use AI for Multilingual Depression Screening

IIT Kharagpur will lead the technical development, including AI model architecture and system integration, while NIMHANS and LGBRIMH will contribute clinical expertise and validation.

NIMHANS, in collaboration with IIT Kharagpur and LGBRIMH Tezpur, has launched HEADS, a two-year project that will evaluate AI-assisted approaches for early depression screening.

Supported by the Wellcome Trust, the initiative will focus on multilingual AI and human oversight in mental healthcare.

The HEADS (Human-in-the-loop Evaluation of Assisted Depression Screening) project has brought together clinical, technical, and regional mental health expertise to explore how AI can support earlier identification of depression.

The initiative will focus on building tools suited to India's linguistic diversity rather than relying primarily on English-language datasets.

Under HEADS, AI models will be developed and evaluated for Kannada, Assamese, Hindi, Bengali, and English. The project is designed around a human-in-the-loop approach, meaning the AI system will support clinicians rather than operate as an autonomous diagnostic tool.

This framework is intended to retain clinical judgement and strengthen safeguards around AI use in mental healthcare.

The project will also involve a 15-member panel of Lived Experience Experts, comprising people with personal experience of mental health conditions.

Their participation will extend across areas including system design, consent protocols, language evaluation, bias audits, and stress-testing of the AI system.

Over 24 months, HEADS is expected to conduct around 4,500 clinical interviews across NIMHANS and LGBRIMH sites.

IIT Kharagpur will lead the technical development, including AI model architecture and system integration, while NIMHANS and LGBRIMH will contribute clinical expertise and validation.

The multilingual approach also addresses a key challenge in mental health AI: regional expressions of distress can carry cultural and linguistic nuances that may not be captured by systems trained predominantly on English data.

The project will therefore need to address issues around dataset availability, cultural interpretation, algorithmic bias, privacy, informed consent, and clinical validation.

As AI adoption expands across healthcare, HEADS could provide evidence on how AI-assisted screening can be evaluated within clinical settings while keeping human oversight central to decision-making.

Stay tuned for more such updates on Digital Health News