Development of a Contactless Patient Fall Detection System Using Radar Sensors

Abstract

Falls are among the most significant causes of injury, functional decline, and loss of independence in older and hospitalized patients. Delayed recognition of an event increases the risk of a prolonged lie, hypothermia, dehydration, rhabdomyolysis, and delayed care for fractures or traumatic brain injury. Wearable sensors, video cameras, and ambient sensors partly address the monitoring challenge but are limited by user adherence, privacy concerns, dependence on lighting, and incomplete room coverage. Radar sensors enable contactless measurement of range, radial velocity, direction, and the spatial distribution of reflections without producing a conventional image. This review examines continuous-wave Doppler, FMCW, UWB, and millimeter-wave MIMO radars, micro-Doppler analysis, range-Doppler maps, point clouds, and machine-learning algorithms. A fall is represented as a sequence comprising loss of balance, reduction in body height, contact with a surface, and the post-fall state. Experimental studies demonstrate high accuracy under controlled conditions; however, clinical generalizability is limited by small samples, the predominance of simulated falls, hardware heterogeneity, and the absence of standardized operational metrics. Implementation requires prospective multicenter studies, participant- and room-level data separation, assessment of false alarms per patient-day, local processing, cybersecurity, and integration with a clinical response workflow. A radar system should be regarded as a component of comprehensive patient safety rather than a substitute for fall prevention and clinical observation.