Zihan Xu bio photo

Email

Github

ThreadStates: State-based Visual Analysis of Disease Progression

Lastest Update: 1st April 2024

Introduction

Explore disease progression status in longitudinal patient cohort data and reveal the association between disease progression and other variables.

image-20240504120815987


Motivation

  • Capabilities in modeling disease progression are not fully utilized
  • The large number of observations, the complex correlation between observations, and the existence of irrelevant observations
  • A disease can exhibit different progression patterns within a target patient cohort
  • The identified disease progression patterns need to be associated with other variables


image-20240504120935108

Highlights

  • Dimension reduction and clustering image-20240504121015503
  • A novel glyph design to depict the feature distribution image-20240504121059438
  • more details about selected state transitions image-20240504132810657
  • Address the false sense of transitions between multiple timepoints caused by Sankey-based visualizations image-20240504132823225

Critical thinking

  • Incorporate domain knowledge into the model