A software engineering framework for designing, governing, and deploying AI-driven, culturally adaptive systems: a mental health case study in Bangladesh

Abstract

Mental health disorders represent a critical and growing public-health challenge in Bangladesh, yet access to affordable, culturally appropriate, and trust worthy psychological support remains severely limited. Empirical findings based on the DASS-21 instrument demonstrate significant levels of depression, anxiety, and stress among Bangladeshi populations, while existing services are constrained by shortages of professionals, linguistic barriers, stigma, and fragmented digital infrastructure. This thesis addresses this gap by proposing a bilingual, culturally adaptive, AI-assisted mental-health screening system designed specifically for low-resource environments. The research adopts a multidisciplinary software-engineering perspective, arguing that the success of AI in mental healthcare depends not only on model accuracy but also on governance, development process, architecture, quality assurance, and deployment economics.