Ensure your working directory is PrediccionGlucosa/casos. For some reason, things don't work otherwise.
Build (and use) a pip3 environment that adheres to the given requirements.txt file.
Run python3 main.py --cn 0.
The dataset and processed files are in D1NAMO.
The code and graphs are in casos.
The models in tf format are in models. There are several cases. Caso_0: There are independent patients. Any of them has 7 experiments (trynumber). There are 6 patients * 7 experiments = 42 models. They are generated several times, 1 for each execution. In total 420 models. Maybe it is posible to train them just once, and then save and load, and use them several times for 10 executions.