Parkinson's Disease Detection from Spiral and Wave Drawings Using CNN - RNN With GSM Alert

This paper detects Parkinson's Disease (PD) using a hybrid model of CNN and RNN when analyzing spiral and wave drawings. The model incorporates both spatial and temporal pattern recognition, which contributes to a significantly improved diagnostic accuracy for this condition. The CNN represents the spatial features, such as edge detection and stroke thickness, whereas the RNN captures temporal dependencies in the sequences of drawings that reflect the motor impairments associated with PD. This non-invasive objective approach also supports early detection which can then be intervened upon promptly. The model's performance was promising in accuracy, precision, recall, and F1-score values. Diagnostic information is transmitted quickly to ensure that an embedded GSM alert system communicates PD status to patients.