DPVis-Waterfall is a Flask application for presenting disease progression pathways. It visualizes sequences of hidden states, learned by Hidden Markov Models, over time on longitudinal observational data. We used the source code to visualize figures in the paper published by Nature Communications:
Progression of type 1 diabetes from latency to symptomatic disease is predicted by distinct autoimmune trajectories,
Bum Chul Kwon, Vibha Anand, Peter Achenbach, Jessica L Dunne, William Hagopian, Jianying Hu, Eileen Koski, Åke Lernmark, Markus Lundgren, Kenney Ng, Jorma Toppari, Jorma Veijola, Brigitte I Frohnert, the T1DI Study Group
Nature Communications, 2022
The full-fledge visualization system was introduced in the following paper published by IEEE TVCG.
DPVis: Visual Analytics with Hidden Markov Models for Disease Progression Pathways,
Bum Chul Kwon, Vibha Anand, Kristen A Severson, Soumya Ghosh, Zhaonan Sun, Brigitte I Frohnert, Markus Lundgren, Kenney Ng
IEEE Transactions on Visualization and Computer Graphics (TVCG), 2020 arXiv preprint (arXiv 1904.11652)
Python 3.7.4
Use the package manager pip to install packages.
pip install -r requirements.txtRun Flask Server on the Project Root Directory.
python app.pyWhile the server is running, launch a modern web browser (e.g., Firefox) and go to https://localhost:4848. It may take 1-3 minutes to completely load the page depending on the computing environment.
For demonstration, we use a synthetic dataset: /static/data/dpvis-csv-file.csv