Finding the Most Common Treatment Trajectories of Patients Based on the DTW Method
Name
Brandon Loorits
Abstract
The goal of this thesis is to produce a workflow to help user to find most common treatment trajectories of some specific disease patients in cohort. Proposed workflow consists of 7 parts - converting the data to the required shape, calculating the similarity matrix by the method of dynamic time warping, clustering, silhouette analysis, correction of trajectories, creation and visualization of result trajectories. The proposed workflow in the thesis potentially helps the user to find the most common treatment trajectories automatically. This workflow uses dynamic time warping to determine the similarity of treatment trajectories, hierarchical agglomerative clustering as the clustering method, and silhouette analysis to evaluate clustering. Workflow results are visualized and printed as output as well.
Graduation Thesis language
Estonian
Graduation Thesis type
Bachelor - Computer Science
Supervisor(s)
Raivo Kolde, Markus Haug
Defence year
2022