Modeling frailty is a key issue for better addressing the needs of aging people. In our research, we introduce a multidimensional frailty model to quantify the level of frailty of aging people in Ambient Assisted Living solutions. The proposed frailty model takes into account three dimensions: activities of daily living, physical, and social. The proposed model has been harmonized with the existing qualitative medical assessments and the International Classification of Functioning, Disability, and Health (IFC). The model has been tested through the deployment of UbiSMART (an AAL framework) during 16 months in 8 real scenarios of aging in place (homes and nursing homes). In this paper, we present the process to convert sensor data into numerical frailty values: the definition of the frailty items, the technical design, the implementation, and the numerical treatment of the data. We also present the evaluation of the model and the results obtained during the deployments.