ABSTRACT People living with dementia (PLWD) from racial-ethnic minoritized groups and socioeconomically disadvantaged environments are more likely to face barriers to diagnosis, care, and services. Multiple social determinants of health (SDoH) contribute to the disparities in Alzheimer’s disease (AD) and AD-related dementias (AD/ADRD) progression and the quality of AD/ADRD care. Thus, AD/ADRD is a public health crisis that must be managed not only by traditional medical care but also by addressing patients’ unmet social needs. Artificial intelligence (AI) and large real-world data (RWD), such as electronic health records (EHR), offer an opportunity to develop innovative approaches that improve health and health equity by addressing SDoH. The objective of this project is to develop a machine learning (ML)-based social risk management platform - ISMART (intelligent Social risk Management in AD/ADRD paTients) - that can be embedded into EHR systems to improve the quality of care and quality of life of PLWD. We will use RWD from the OneFlorida+ network, a member of the National Patient-Centered Clinical Research Network (PCORnet), comprising EHR data from >20M individuals. We will leverage our prior work that established an external exposome database with contextual SDoH measures documenting social and physical environments and a natural language processing pipeline that can extract person-level SDoH (including caregiver information) from clinical narratives in EHRs. Our study will follow an intervention mapping approach that engages a Stakeholder Advisory Committee to achieve three Specific Aims. In Aim 1, we will build an RWD cohort of PLWD and to identify key contextual and person-level SDoH associated with PLWD care and outcomes. In Aim 2, we will develop ML- based social risk management algorithms for dementia care and outcomes, including (a) a fair individualized polysocial risk score (iPsRS) to screen for unmet social needs in PLWD; and (b) causal-principled AI methods to quantify the causal, heterogeneous effect of key actionable SDoH (e.g., food) on PLWD care and outcomes. In Aim 3, we will co-design with stakeholders the ISMART platform, including (a) prototyping ISMART platform following a User-Centered Design process; and (b) developing recommendations for future implementation and evaluation via focus groups and Delphi panels. The success of our project will lead to the development of ISMART prototype for social risk management in PLWD, with a set of strategies for future implementation and evaluation. Our innovative, structured approach to integrating social risk management with health care of PLWD may lead to a necessary paradigm shift in US health care delivery.