The population of older adults is increasing across the globe; this growth is predicted to continue into the future. Most older adults prefer to live in their own home, but many live alone without immediate support. Living longer is often coupled with health and social problems and difficulty managing daily activities. Therefore, some level of care is required, but this is costly. Technological solutions may help to mitigate these problems by recognising subtle changes early and intervening before problems become unmanageable. Understanding a person's usual behaviour when carrying out Activities of Daily Living (ADL) makes it possible to detect and respond to anomalies. However, current commercial and research monitoring systems do not an analysis of ADL and are unable to detect subtle changes. To address this gap, we propose the STRETCH (Socio-Technical Resilience for Enhancing Targeted Community Healthcare) sensor platform that is comprised of non-invasive sensors and machine learning techniques to recognise changes and allow early interventions. The paper discusses design principles, modalities, system architecture, and sensor network architecture.