Two possible versions are included: final and raw versions.
The first one (FINAL_VERSION.rar) contains all the data that results from running the application.
The second (RAW_VERSION.rar) only contains the original raw data (available in MIMIC-II/Physionet).


___________________________________________________________________
1------------FINAL VERSION ----------------------------------------
иииииииииииииииииииииииииииииииииииииииииииииииииииииииииииииииииии
This version contains all the data resulting from the steps (1, 2, 3, 4 and 5), to be explained bellow (section 2.3 Ц Raw version).


1.1 ии Content иииииииииииииииииииииииииииииииииииииииииииииииииииииии
When decompressing this file FINAL_VERSION.rar the following folders will be create:
\codeMatlab: matlab code
\MIMIC_TRAIN_TXT:     MIMIC II train datasets H and C
\MIMIC_TEST_MAT:     MIMIC II test datasets A and B
\dataTrain:     Preprocess H and C datasets
\dataTest    :     Preprocess A and B datasets
\NNmodels:     GRNN models, resulting from the training process
\Results:        Predictions and occurrences of AHE episodes


1.2 ии Run the application ииииииииииииииииииииииииииииииииииииииииииииииииииииии
This is a Matlab program. 
MATLAB Version 7.5.0.342 (R2007b)
Operating System: Microsoft Windows Vista Version 6.0 (Build 6000)
Java VM Version: Java 1.6.0 with Sun Microsystems Inc. 

Moreover, it is assumed that signal processing and neural networks toolboxes have been installed.
To run the application, in the folder codeMatlab, run the m file
    main_AHE or makeFile
Five options will be shown, each one corresponding to one of the necessary processing phases.



____________________________________________________________________
2  ----------------- RAWVERSION --------------------------------------
ииииииииииииииииииииииииииииииииииииииииииииииииииииииииииииииииииии
This version only contains the available data in MIMIC-II/Physionet.

2.1 ии Content иииииииииииииииииииииииииииииииииииииииииииииииииииииии
When decompressing RAW_VERSION.rsr file the following folders will be create:
\codeMatlab:    matlab code
\MIMIC_TRAIN_TXT:    MIMIC II train datasets H and C
\MIMIC_TEST_MAT:    MIMIC II test datasets A and B
\dataTrain:    Empty: will contain the preprocess H and C datasets
\dataTest    :    Empty: will contain the preprocess A and B datasets
\NNmodels:    Empty: will contain the GRNN models, resulting from training \Results:        Empty: will contain the occurrences of AHE episodes


2.2 ии Run the application ииииииииииииииииииииииииииииииииииииииииииииииииииииии
This is a Matlab program. Moreover, it is assumed that signal processing and neural networks toolboxes have been installed.
To start the program, in the folder codeMatlab, run the m file
    main_AHE or makeFile


2.3 ии Steps иииииииииииииииииииииииииииииииииииииииииииииииииииииии
Basically, the application consists of five steps (that should be executed in the order that they are defined):

[1] -------------
Pre Processing ABP signal: (mainPreProcessingABPsignal.m matlab file)
This step gets the raw data and passes it through a set of pre-processing techniques, namely to deal with missing information, resampling and noise reduction.
The train signals (H,C) are placed in the folder \dataTrain. The test signals (A,B) in the folder \dataTest.
     \MIMIC_TRAIN_TXT\{H,C}; --> \dataTrain\{H,C}
    \MIMIC_TEST_MAT\{A,B}   --> \dataTest\{A,B}

[2] -------------
Train ABP Signals (mainTrainABPsignal.m matlab file)
This step gets the preprocessed H and C ABP signals and generates the respective GRNN models.
The resulting models are placed on \NNmodels folder.
      \dataTrain --> \NNModels

[3] -------------
Merge HC (mainMergeHC.m matlab file)
This step gets the preprocessed H and C ABP signals and generates the template dataset. The result is placed on \dataTrain
     \dataTrain --> \dataTrain

[4] -------------
Forecast ABP signal (mainForecastABPSignal.m matlab file)
This step gets the preprocessed A and B ABP signals, as well as the previous trained GRNN models and generates the ABP signal forecast.
Based on the predictions signal the occurrence of a possible AHE is verified.
The resulting prediction signals as well as the occurrences of AHE models are placed on folder \Results
      \dataTest + \NNmodels    -->  \Results

[5] -------------
Show the results (mainShowAHE.m matlab file)
Finally, with this option it is possible to see the results generated in step 4.
      \Results --> AHE_ {A,B } visualization