Parkinson’s disease detection through hand drawings using Alexnet model

Innovative methods based on deep learning and different types of data are very useful in diagnosing Parkinson's disease.Through this research, we present a new method for detecting Parkinson's disease using deep convolutional neural networks based on the AlexNet architecture.The proposed approach focuses on using hand drawings of injured or suspected people and then integrating the features extracted from these drawings to classify subjects.To evaluate the proposed method, we used spiral and wave patterns available in the Kaggle's repository.This database is obtained from patients with Parkinson's disease and healthy individuals.By combining the features of the two different hand drawing styles, we were able to significantly improve detection accuracy.The related experimental results show the effectiveness of this approach, as it achieved an accuracy of 96.67% on the test set, outperforming current methods that use fusion in classifiers to make decisions, which achieved an accuracy of 93.33%.However, the challenge of achieving a higher accuracy highlights the complexity of diagnosing Parkinson's disease from manual drawings.