-
In this part we need to preprocess the data.
Acquire physiomarkers that are important for identifying the sepsis disease. -
Since the data from CITI website can't be accessed due to unavailabitlity of login credential. I can't find a good high frequency of sepsis dataset online thus we need to skip this part for now.
-
Lets assume we have the data set with all the physiological data
[temperature, sao2, heartrate, respiration, cvp, etco2, st1, st2, st3, icp] -
We will use various statistics method to get good physiomarkers for the dataset. Lets assume these are the important physiomarkers for sepsis prediction [temperature, sao2, cvp, etco2, systemicystolic, systemicmean, st2,st2, icp]
-
In this part we will make different Temporal convolution neural network (t-CNN) architecture with input data as the physiomarkers that are acquired in phase 1.
-
Training the architectures on train data.
-
Testing the architectures on test data.
-
Evaluating those architectures on the different metrics to identify the best performing architecture.
-
In this part we will Integrate the model to a python module.
-
Unit test the module created.
-
Add all the documentation for further enhancement of the project in future