Multimodal Diagnostic System For Detecting Nerve Weakness Through Symptom Analysis: Combining Machine Learning And Rule-Based Approaches

Nerve weakness, often marked by symptoms such as numbness, muscle stiffness, and memory loss, poses significant diagnostic challenges due to its diverse manifestations. This study presents a multimodal diagnostic approach that merges rule-based methodologies with advanced machine learning and deep learning techniques to improve the identification accuracy of nerve-related conditions. The primary goal is to build a robust rule-based model that links specific symptoms to potential neurological issues using a comprehensive dataset. The second objective is to enhance this model by integrating multiple data inputs and utilizing machine learning and deep learning algorithms to boost diagnostic precision. The third aim is to support early diagnosis, treatment, and potential prevention of neurological and systemic diseases, including Multiple Sclerosis (MS), Parkinson’s Disease, Alzheimer’s Disease, and related disorders. The dataset includes over 1,000 patient records from various regions of Punjab and other states in India, containing details such as symptom onset and specific neurological symptoms like muscle weakness, headaches, and memory loss. Data was gathered through clinical interviews, video/audio recordings, wearable sensors, and diagnostic procedures such as Electromyography (EMG), MRI, and EEG. Machine learning and deep learning models, including Convolutional Neural Networks (CNNs), were employed to analyze the data. The hybrid system achieved diagnostic accuracy between 91% and 97%, significantly outperforming the standalone rule-based system’s 82% to 92% accuracy. In the final phase, XGBoost (Extreme Gradient Boosting) was used to further refine model accuracy by using gradient boosting techniques, improving predictive capabilities.