GENSCORE Project: Leveraging Generative AI to adapt and enhance the scoring of widely used mental health assessment tools in low-resource context

In Nigeria where about 250 psychiatrists serve over 220 million people, contextually accurate mental health screening tools are not just beneficial, they are essential. Our GENSCORE Project aims to develop and validate a culturally grounded, multimodal AI platform/software to improve depression and anxiety screening for Pidgin English, Hausa, and Yoruba speakers in Nigeria. The project will build on our successful pilot to co-develop GENSCORE-DATA v2.0, one of Africa’s largest multilingual mental health dataset, and fine-tune GENSCORE-LLM v2.0 for culturally aligned interpretation. Our culturally relevant model, GENSCORE-LLM, will enable the development of contextually relevant mental health screening (GenPHQ and GenGAD) tools co-designed with people with lived experience and mental health experts. These tools will be integrated into an open-access, multimodal GENSCORE platform combining voice (audio) and self-reported text, facilitating real-time screening and decision support. The GENSCORE tools will undergo rigorous community and clinical validation and benchmarked against SCID-5-CV assessments among 990 participants in language-specific sites: Lagos State (Pidgin English), Abuja, FCT (Hausa) and Oyo State (Yoruba). Robust ethical, governance, and safety frameworks will ensure fairness, data sovereignty, and responsible AI use, with broad national and international dissemination.