Multimodal sensing and machine learning for continuous monitoring of agitation in dementia

Multimodal sensing combined with machine learning techniques, could allow for more continuous, objective and quantifiable monitoring of agitation in dementia. This could lead to a better evaluation of interventions, lower need for potentially dangerous psychotropic drug use, less caregiver stress and lower economic costs. Our results support the initial hypothesis of population heterogeneity and they represent the base for personalized agitation detection and treatment evaluation using physiological markers and environmental triggers.