Research on the Application of Improved Particle Swarm Optimization Algorithm Assisted Multi-feature EEG Data Processing in the Diagnosis of Frontotemporal Dementia

Frontotemporal dementia (FTD) seriously impairs the functions of the nervous system, and its early symptoms are concealed, making it difficult for traditional diagnostic methods to accurately identify it in the early stage. This study focuses on the processing of electroencephalogram (EEG) data, aiming to assist in the diagnosis of FTD.Firstly, 15 features are extracted by integrating the dimensions of entropy, time-frequency domain and STA to analyze the EEG data. Feature selection is carried out through Pearson correlation analysis and SHAP interpretability analysis to obtain a simplified feature subset. In view of the deficiencies of the particle swarm optimization (PSO) algorithm in feature selection, an improved IMPSO algorithm is proposed. The sine chaotic mapping initialization and the dynamic particle position update equation are introduced to enhance the ability to search for the optimal feature subset in high-dimensional space. Through the IMPSO algorithm to screen the EEG channel features, 11 key channels are determined and input into the XGBoost classification model, achieving an accuracy rate of 88.86%.This study provides a new data processing method for the diagnosis of FTD and has the application potential in large-scale EEG data processing and resource-constrained scenarios.