Neurodegenerative diseases detection and grading using gait dynamics.

Area of Science:

• Neurology
• Biomedical Engineering
• Artificial Intelligence

Background:

• Neurodegenerative diseases (NDs) like Parkinson's, Huntington's, and ALS pose significant clinical challenges.
• Gait analysis offers a simple, non-invasive method for assessing neurological function.
• Accurate detection and severity grading of NDs are crucial for effective patient management.

Purpose of the Study:

• To develop an AI-based system for detecting and predicting the severity of neurodegenerative diseases using gait features.
• To differentiate between Parkinson's disease, Huntington's disease, Amyotrophic Lateral Sclerosis, and control groups.
• To assess the performance of various machine and deep learning models for ND detection and severity grading.

Main Methods:

• Utilized gait signals to extract relevant features for AI analysis.
• Developed classification models for disease detection (4-class and disease vs. control subgroups).
• Employed machine and deep learning techniques for disease severity prediction within specific disease subgroups.

Main Results:

• Achieved measurable performance in disease detection using metrics like Accuracy, F1 Score, Precision, and Recall.
• Quantified prediction performance for disease severity grading using metrics such as R, R², MAE, MedAE, MSE, and RMSE.
• Demonstrated the potential of AI-driven gait analysis for ND diagnosis and severity assessment.

Conclusions:

• AI-powered gait analysis is a viable and non-invasive approach for detecting and grading neurodegenerative diseases.
• The developed system shows potential for clinical application in early diagnosis and monitoring of NDs.
• Further research can refine these models for improved accuracy and broader clinical utility.