Validation of Artificial Intelligence-Based Facial Paralysis Assessment in Patients With Bell's Palsy
Validation of Artificial Intelligence-Based Facial Paralysis Assessment in Patients With Bell's Palsy
This observational study aims to assess the concurrent validity of an artificial intelligence (AI)-based facial paralysis assessment system in patients with unilateral Bell's palsy. Currently, clinical assessment relies on subjective scales like the Sunnybrook Facial Grading System, which can vary between different observers. This study will compare AI-generated composite asymmetry scores-derived from real-time computer vision analysis of facial landmarks-with scores from the Sunnybrook system. The goal is to determine if AI can provide a valid, objective method for monitoring facial nerve recovery.
Participants with unilateral Bell's palsy will be recruited for a single assessment session. The assessment involves two primary components:
Clinical Assessment: A qualified physical therapist will grade the patient's facial function using the Sunnybrook Facial Grading System, which evaluates resting symmetry, degree of voluntary movement, and synkinesis.
AI Assessment: A computer-vision-based system will utilize a standard camera to detect 468 facial landmarks in real-time. The system calculates a composite asymmetry score by comparing the movement amplitude and positioning of the affected side of the face against the healthy side during standardized facial expressions.
The study will utilize Spearman's rank correlation coefficient to analyze the relationship between the AI-derived scores and the Sunnybrook scores to establish concurrent validity. No personal images or videos will be stored; the AI performs real-time processing and immediate data deletion to ensure participant privacy.
Inclusion Criteria:
Exclusion Criteria:
alionour22@gmail.com+201142154162