Proactive Agitation Detection in Dementia: An AI-Powered Multimodal Approach for Caregiver Support
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
Agitation in Alzheimer's patients creates serious problems for caregivers. When agitation escalates, patients may exhibit aggressive behaviors such as hitting or yelling, often necessitating sedation to de-escalate. Currently, caregivers rely on subjective observation, frequently missing early warning signs. By the time agitation is noticed, it's usually too late to prevent an aggressive incident. Our project develops an AI-based multimodal system to automatically detect agitation in a patient and support timely intervention decisions. Our project develops an AI-based multimodal system to automatically detect agitation in a patient and support timely intervention decisions. Wearable devices track subtle physiological signs that typically appear before agitation escalates, monitoring movement patterns, heart rate changes, and sleep quality. The system also incorporates facial emotion recognition, body movement analysis, and audio-verbal monitoring to capture a fuller picture of the patient’s state. When concerning trends are detected across these combined signals, caregivers are alerted in time to intervene. With advance warning, caregivers can step in early by adjusting the environment, redirecting the patient's attention, or administering medication before escalation occurs. The goal is to prevent aggressive episodes rather than just responding to them. Early intervention could reduce the number of violent incidents and cut down on how often patients need sedation. For caregivers, this means fewer injuries, less stress, and more confidence in managing difficult situations. Patients benefit too, receiving calmer, more dignified care instead of emergency interventions. This shift from reacting to crises to preventing them could improve outcomes for everyone involved in Alzheimer's care.
Fields of Study & Subjects
Dementia and Cognitive Impairment Research
Context-Aware Activity Recognition Systems
Emotion and Mood Recognition
Publication details
- Journal / source
- ODU Digital Commons (Old Dominion University)
- Publication date
- 25 March 2026
- Language
- EN