Cluster analysis of temporal trajectories of hospital laboratory examinations
Shoji Hirano, Shusaku Tsumoto · Proceedings of SPIE, the International Society for Optical Engineering/Proceedings of SPIE · 2007
This paper proposes a new approach to temporal trajectory analysis for clinical laboratory examinations. When we select m laboratory examinations, their temporal evolution for one patient can be viewed as a trajectory in m-dimensional space. Multiscale comparison technique can be applied for segmentation and calculation of structural similarities of such trajectories. Then, clustering cna be applied to the calculated similarities for classiffication of these trajectories. The proposed method was evaluated on hepatitis datasets, whose results show that the clustering captured several interesting patterns for severe chronic hepatitis.