To provide insights into the application of facial expression recognition (FER) in stroke identification and rehabilitation monitoring, highlighting its potential impact on patient outcomes.
Key Findings:
A total of 1,855 studies were identified, of which nine met inclusion criteria, including eight diagnostic studies and one rehabilitation trial.
FER demonstrated diagnostic utility for stroke with accuracies ranging from 82% to 98%.
Specific tasks like KISS and SPREAD were particularly effective in assessing facial asymmetry.
One study achieved 99.81% accuracy in monitoring rehabilitation intensity through real-time facial expression classification.
Interpretation:
FER technology shows significant potential as an auxiliary tool for stroke diagnosis and rehabilitation, enabling precise analysis of facial movements, which could enhance clinical decision-making.
Limitations:
FER models face challenges in real-world clinical translation, including integration into existing workflows and variability in patient populations.
Current studies are limited in number and scope, necessitating further research.
Conclusion:
Future research should integrate multimodal data, such as neuroimaging and patient-reported outcomes, and utilize real-world databases to enhance the clinical implementation of FER technology, improving care delivery and reducing patient mortality and disability.
A nationwide Korean cohort study found modestly higher Parkinson's disease incidence among patients with asthma or allergic rhinitis, with stronger associations observed among patients with greater allergic disease burden.
