k-Prototype Clustering

Cluster quality is evaluated using silhouette coefficient, Elbow method, UMAP visualisation, and Adjusted Rand Index stability checks. k-prototype clustering handles mixed numerical and categorical data using Euclidean and Hamming distance…

1 sources - 4 claims

Cluster quality is evaluated using silhouette coefficient, Elbow method, UMAP visualisation, and Adjusted Rand Index stability checks. k-prototype clustering handles mixed numerical and categorical data using Euclidean and Hamming distances. The k-prototype algorithm stratifies patients using natural, health, and behavioural attributes. Biomarkers are included because they can reflect stroke clinical status or evolution and complement clinical endpoints.