A new spatial millimetre-wave (mm-wave) radar point cloud dataset has been introduced for health applications, specifically targeting human activity recognition (HAR) and pose estimation in remote patient monitoring, personalised care, and home-based rehabilitation. The dataset was developed as a controlled healthy-reference resource, providing a standardised baseline for early-stage algorithm development rather than targeting unconstrained clinical rehabilitation sessions. It includes both unprocessed original radar data and pre-processed radar point clouds, allowing researchers to either use the existing baseline or develop custom signal-processing pipelines. The dataset was collected under diverse lighting conditions, addressing a gap in existing datasets that often focus on general daily activities or mainly involve young adult participants. Unlike many existing datasets that provide only processed radar point clouds, this dataset includes both original data and processed point clouds. The background notes that mm-wave radar is a non-contact sensing modality that does not capture optical images and can operate under different lighting conditions, making it relevant for research on home-based and rehabilitation-oriented motion analysis. The dataset aims to support movements involving reaching, balance control, posture transition, and limb coordination, which are often required in rehabilitation-oriented studies.