The 'tpp.py' program goes through the JSON files in the given fake hospital patients dataset and reorganises it into an easy to use dictionary for each patient. This includes sequencing all the observations over time a patient has had from different visits to the hospital, this would be difficult to generalise for patients with a different number of observations if inputting into a neural network. To combat this I chose to break this list into 3 features: average reading, gradient and range (these can easily be adapted depending on what the statitician would need/want to investigate).

The 'analysis.py' file guides you through how to build datasets, visualise and export to csv.

The 'ann-classifier.py' file guides you how to run an artificial neural network on a dataset selecting the column you want to use as your classifier.

• Extract 'fhir.7z' into the diectory with the files tpp.py and analysis.py
• From terminal run 'python tpp.py', this extracts all the features specified in the file into a more useable format (described above).
• Perform analysis:
  • Run 'python analysis.py', this walks you through how to use the interface.

The analysis file can visualise and create datasets and then export the dataset when you're done into csv format

• Run ANN:
  • Run 'python ann-classifier.py', this walks you through how to run a neural network.