Determining gait patterns and abnormalities of a user includes forming a plurality of point clouds corresponding to the user, each of the point clouds being three-dimensional coordinates of moving points, frame by frame, through a data capturing session, determining centroids of the point clouds, determining momentary walking velocities using estimates based on vectors connecting the centroids for adjacent frames captured during walking of the user, determining gait speed for the user based on the momentary walking velocities, determining at least one distribution of gait speeds for the user, and detecting gait abnormalities based on deviation of the gait speed from the at least one distribution of gait speeds. Detecting a plurality of point clouds may include using a tracking device to capture movements of the user. The tracking device may use radar and/or lidar. The system may determine a gait pattern of the user corresponding to routines of the user.