Deep Learning-Based Evaluation of Postural Control Impairments Caused by Stroke Under Altered Sensory Conditions

Area of Science:

• Neuroscience
• Biomedical Engineering
• Artificial Intelligence

Background:

• Postural control impairments are common in stroke survivors, increasing fall risk.
• Current clinical assessments lack the precision to capture complex postural deficits.
• Advanced computational methods are needed for accurate diagnosis and rehabilitation.

Purpose of the Study:

• To develop and validate a hybrid deep learning framework for classifying sensory dysfunction in stroke patients.
• To improve the accuracy and robustness of postural control assessment.
• To support personalized rehabilitation planning and fall prevention strategies.

Main Methods:

• A hybrid deep learning model combining Convolutional Neural Networks (CNNs) and Type-2 fuzzy logic was developed.
• The model was trained and tested on a large dataset (8316 samples) from stroke patients using EquiTest data.
• Performance was evaluated under various sensory manipulation conditions and simulated measurement noise.

Main Results:

• The proposed framework achieved high classification performance: 97% accuracy, 96% precision, 97% sensitivity, and 96% specificity.
• It outperformed traditional CNNs and other baseline classifiers in identifying sensory dysfunction.
• The system demonstrated robustness to noise, performing reliably in simulated clinical settings (down to 1 dB SNR).

Conclusions:

• The hybrid deep learning system offers a practical, non-invasive tool for diagnosing postural control impairments in stroke patients.
• This data-driven approach can significantly aid in personalized rehabilitation planning and fall prevention.
• The findings support the integration of advanced AI in clinical decision-making for stroke care.