The invention relates to the technical field of intelligent medical auxiliary diagnosis and
neural function automatic evaluation, in particular to an NIHSS
score automatic
evaluation system for a
stroke patient. Comprising a multi-mode induction and
perception unit which is used as a front-end
data entry and is used for collecting
patient response in real time to generate a video
stream containing depth and color information and a synchronous audio
stream; the dynamic reference calibration unit is used for extracting kinematic characteristics to construct an individualized nerve reference template; the neural motion spectrum
decomposition unit is used for generating a spectrum
pathological feature vector for distinguishing myasthenia and
ataxia; the opposite-side image rejection analysis unit is used for generating compensation and driving confidence for representing a real nerve driving intention; the cross-
modal reflection analysis unit is used for generating a sensory pathway integrity index according to the
nerve conduction velocity difference; and the collaborative scoring decision engine is used for mapping the multi-
modal features into standardized NIHSS scores. According to the method, the interference of age and basic physique on scoring is effectively eliminated, and a high-precision comparison reference can be provided for subsequent abnormal judgment of the affected side.