부산대학교 도서관

Early detection of AD is crucial for effective intervention and management of the disease. However, existing diagnostic methods are often costly, time-consuming, and reliant on the presence of noticeable clinical symptoms. To address these limitations, there is increasing interest in developing novel screening methods for early-stage AD detection. Reservoir Computing (RC), a machine learning technique inspired by the brain's computational principles, offers a promising approach for AD screening. In particular, the use of Reservoir Computing models, such as Echo State Networks (ESNs), has shown remarkable potential in various cognitive tasks, including pattern recognition and time series prediction. One of the key advantages of RC-based approaches is their ability to learn from unlabelled or weakly labelled data, enabling the detection of subtle changes associated with AD progression. The unsupervised learning aspect of RC models allows them to uncover hidden patterns indicative of AD even before clinical symptoms manifest. Overall, RC-based screening for early-stage Alzheimer's detection holds great promise as a non-invasive, cost-effective, and efficient approach. By harnessing the temporal dynamics of brain activity, RC models can extract informative features from EEG and other temporal data sources. The unsupervised learning capabilities of RC models enable the detection of subtle changes associated with AD pathology, even in the absence of overt symptoms. The computational efficiency and real-time processing of RC models facilitate their practical implementation in clinical settings. Further research should focus on large-scale clinical validation and the development of user-friendly interfaces to translate RC-based AD screening into effective diagnostic tools for early intervention and improved patient outcomes.Keywords---Reseroir computing,