2026-05-10T01:21:32.291Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:10800 2024-12-05T19:30:44Z 3

Exploring Echo State Network for Detection of Gait Freezing in Parkinson’s Patients Optimized Through Modified Metaheuristics, Chapter in LNEE Lecture Notes in Electrical Engineering: PEIS2024 2024: Power Engineering and Intelligent Systems, Springer, volume 1247 2024 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/10800 https://link.springer.com/chapter/10.1007/978-981-97-6714-4_5 N. Bacanin L. Jovanovic G. Radic M. Zivkovic K. Kumpf M. Antonijevic Parkinson’s disease (PD) is a persistent and slowly progressing neurodegenerative disorder that primarily affects movement. PD is the second most prevalent neurodegenerative condition; symptoms typically appear years before a patient is diagnosed, and treatment is generally delayed until symptoms become severe and interfere with everyday life. As there is no known treatment to revert neural damage, preventative treatments are crucial. Early diagnosis and detection, therefore, play a vital role in improving the quality of life. This work explores the use of artificial intelligence algorithms to tackle Parkinson’s detection from time-series data gathered through a non-invasive shoe-mounted sensor array. The diagnostic potential of a recently introduced echo state network (ESN) architecture is explored. Yet, as the effectiveness of this method greatly depends on choosing the right hyperparameters, metaheuristic algorithms are used to maximize performance, and a modified version of the previously suggested crayfish optimization algorithm is presented to further improve performance. The diagnostic approach is tested on a real-world dataset and compared to several contemporary optimizers. The changes demonstrate that the algorithm performed better than the optimizer’s starting version and a number of more recent optimizers that were analyzed, producing the top models with an accuracy higher than 69%. bookPart Springer, Singapore 57 68 10.1007/978-981-97-6714-4_5 LNEE Lecture Notes in Electrical Engineering: PEIS2024 2024: Proceedings of International Conference on Power Engineering and Intelligent Systems, volume 1247