| CPC A61B 5/112 (2013.01) [A61B 5/1127 (2013.01); A61B 5/4082 (2013.01); A61B 5/4088 (2013.01); G06N 20/00 (2019.01); G06T 7/74 (2017.01); G06V 20/46 (2022.01); G06V 40/25 (2022.01); G06T 2207/10016 (2013.01); G06T 2207/20081 (2013.01); G06T 2207/30196 (2013.01); G06T 2207/30204 (2013.01); G06T 2210/22 (2013.01)] | 15 Claims |
|
1. A method for gait-based testing for a neurodegenerative condition in a patient, the method comprising:
processing video data of the patient walking to determine gait kinematic data for the patient, the gait kinematic data comprising time-dependent signals representing at least one of joint positions, joint angles, or body-segment rotations;
processing the gait kinematic data, using one or more computer processors, by segmenting the time-dependent signals by stride and deriving, from the segmented time-dependent signals, one or more gait metrics indicative of a degree of variability, between strides, in characteristic shapes of the time-dependent signals, the one or more gait metrics collectively constituting a gait signature associated with the neurodegenerative condition; and
operating a machine-learning model on input comprising the gait signature, using the one or more computer processors, to determine at least one predictive score associated with the neurodegenerative condition and the patient,
wherein the machine-learning model has been trained on a training dataset comprising, for each of a plurality of patients, multiple gait metrics derived from gait kinematic data acquired for the respective patient, paired with one or more evaluation scores, including a neuropathology score derived from a brain scan of the respective patient, that quantify the neurodegenerative condition in the respective patient, and
wherein the one or more gait metrics collectively constituting the gait signature have been selected from the multiple gait metrics in the course of training the machine-learning model by feature reduction.
|