Deep neural networks for the diagnosis of Parkinson's disease through visual feature learning

Amongst many neurodegenerative diseases, Parkinson&s;s disease is also susceptible to motor symptoms and non-motor indications. The indications increase as time without suitable treatment goes on. Diagnosing patients as usual without any modern techniques will not be free from unfair and slow even. This study presents a subtle technique that distinguish the patient suffering from Parkinson&s;s disease or any other disease sharing some common indications. The model combines convolutional layers for capturing spatial features with fully connected layers for decision-making. To check how well everything works, we track accuracy, sensitivity, specificity, and AUC-ROC. Our framework beats traditional machine learning methods and stands up to some of the best models out there. We also dig into how the model makes its choices, highlighting the visual clues it uses to flag the disease. these deep neural networks really help clinicians catch Parkinson&s;s earlier and with more confidence.