To reproduce the results and to get the data essential for the dashboard, run the notebooks in the following order:
- import-cleaning.ipynb
- clustering-for-cdss-evaluation
- evaluate-clustering
- clustering-for-syndromic-surveillance
- explorative-analysis
Comparing clustering based on MODN's representation of patients to traditionally clustering based on features (demographic and symptoms). Does it help against the inevitable structural missingness of CDSS?