Neural network models to predict ulcerative colitis activity using standard clinico-biological parameters
Description
R scripts for predicting ulcerative colitis endoscopic activity through standard clinico-biological parameters using three neural network models are found in the files provided. First binary model to predict active/inactive endoscopic disease using seven categorical and 13 continuous input variables is built and tested in UCDiseaseActivity_1stNNModel.R script. Console outputs for this script are shown in UCDiseaseActivity_1stNNModel_ConsoleOutput.txt. Second binary model to predict active/inactive endoscopic disease using 12 biological input variables is built and tested in UCDiseaseActivity_2ndNNModel.R script. Console outputs for this script are shown in UCDiseaseActivity_2ndNNModel_ConsoleOutput.txt. The multiclass model to predict Mayo endoscopic score using seven categorical and 13 continuous input variables is built and tested in UCDiseaseActivity_3rdNNModel.R script. Console outputs for this script are shown in UCDiseaseActivity_3rdNNModel_ConsoleOutput.txt.