Study on high-dimensional biomedical data mining method based on K-means clustering algorithm

To address the issue of incomplete data mining caused by missing values, this study proposes a high-dimensional biomedical data mining method based on the K-means clustering algorithm. The methodology comprises three main steps: First, missing data values are imputed using the FCMSI algorithm. Second, linear discriminant analysis is employed for feature extraction, where generalised eigenvalues are calculated based on the inverse matrix of the intra-class divergence matrix, enabling dimensionality reduction through projection. Finally, the K-means algorithm is applied for data mining, incorporating probability weights to reselect clustering centres and avoid local optima. Experimental results demonstrate that the proposed method achieves lower mean square error and logarithmic loss while producing more comprehensive data mining outcomes.