SCOPE = SEER Clinical Outcome Prediction Expert. The Excel book provides the information on the input required for SCOPE. The output example provides the information on what the outputs are. SCOPE has 3 keep functions:
1. Simply calculate the Area under the ROC using the provided risk groups as in the Excel sheet for example.
2. Optimize the risk levels by maximizing ROC area, then do 1.
3. T-Test the Areas of ROC obtained from 1 and 2, this tells you if the optimization improvement is statistically significant.
These SCOPE functions rely on the kernel functions that do the binaryfusion, AROC calculations and optimizations. They need to be in the same folder. There are intermediate output files from SCOPE that could be used to study or verify the outputs.
These are the steps to get the output:
1. Run the ScopeVer1 GUI.
2. Load, this initial the arguments needed for the calculations
3. Let the example excel to load (provided).
4. Select the riskfactor interactively from Excel.
5. Select the observedoutcome interactively from Excel.
6. Push AROC, with binaryfusion.
7. Follow the prompts to interactively select the variables from the Excel.
8. SCOPE will output the results including an example ROC plot, the results of Monte Carlo simulation, as well as the final risklevels.
9. Repeat with AROC without optimization.
10. Push the t-test button and select the cutoffROC.mat and perfcurveROC.mat to see if the two samples of ROC areas are statistically different. The cutoffROC.mat and perfcurveROC.mat files are provided for this example. Users may want to test if AROC with versus without the binary fusion.
Rex Cheung (2019). SCOPE Ver 1 GUI (https://www.mathworks.com/matlabcentral/fileexchange/37398-scope-ver-1-gui), MATLAB Central File Exchange. Retrieved .