KOMAP shiny app code

The goal of KOMAP is to fit an unsupervised phenotyping system which does not require manual chart review to acquire gold standard labels for algorithm training and validation. The online algorithm only requires summary-level covariance matrix for coefficient training. To get a quick model evaluation without individual-level data, users can upload conditional mean vector and covariance matrix to obtain a simulated model AUC. KOMAP can deal with cases when both codified and NLP features are available, or when only one feature type is at hand.