BRAINS: A Retrieval-Augmented Agent for Alzheimer's Detection and Monitoring

Md Kishor Morol, Md Tanzib Hosain, Nafiz Fahad, Md. Jakir Hossen, Mohammad Ali Moni

Keywords: Alzheimer’s disease, retrieval augmented generation, cognitive decline detection, clinical decision support.

Registration Requirement: Yes

Abstract: As Alzheimer’s disease (AD) continues to impose a growing global burden, early and accurate detection remains essential, particularly in low-resource settings. To address this challenge, we propose BRAINS (Biomedical Retrieval-Augmented Intelligence for Neurodegeneration Screening), a retrieval-augmented framework for Alzheimer’s detection and monitoring. BRAINS combines a Diagnostic Module, which applies fine-tuned LLMs to cognitive and neuroimaging data such as MMSE, CDR, and brain volume measures, with a Case Retrieval Module that retrieves similar patient profiles from a curated knowledge base. Retrieved cases are integrated through a Case Fusion Layer to improve contextual reasoning before inference. Experiments on real-world datasets show that BRAINS effectively identifies early cognitive decline and classifies disease severity, highlighting its promise as a scalable, explainable tool for early-stage Alzheimer’s screening.

Visa & Travel: Yes

Read CFP & Author Instructions: Yes

Originality Policy: Yes

Single-blind & Not Under Review Elsewhere: Yes

LLM Policy: Yes

Submission Number: 21