closedLoop-CT4TD

This repository contains the source code to replicate the case study presented in A Closed-loop Optimization Framework for Personalized Cancer Therapy Design

Requirements

The source code is written in Python 3.6 and it requires the following packages:

• CASADI (https://web.casadi.org/docs/#document-install)
• numpy
• pandas
• matplotlib
• sklearn
• scipy

Reproducibility

To replicate the case study presented in the article, execute full_cycle.py. At the end of its computation, you will find, for each value of phi, a folder that contains the corresponding results for each patient (i.e. for phi=50 and patient=patient_1 the folder is datasets/simulations/50/patient_1.csv). In each result file you will find the following columns:

1. Number of measured CSCs.
2. Estimated decay (net growth rate) of CSCs (1/day).
3. Estimated value of EC₅₀ (mg/L).
4. Mean concentration of Imatinib in the blood (mg/L).
5. The AUC of the time point under study (mg*hours/L).

New case study

If you wish to generate a new dataset, you need to run createsample_patient.py and then, you can run full_cycle.py to obtain the results. When you run createsample_patient.py, you will create a new file synthetic_patients.csv, which contains a line for each patient with the following columns:

1. Age (years).
2. Body weight (kg).
3. Ground truth EC₅₀ (mg/L).
4. Tumor parameter a₁.
5. Tumor parameter p₁ (1/days).
6. Initial number of CSCs L₁(0).

Analysis of standard dosage

Finally, if you wish to simulate standard Imatinib dosage for each patient, you need to run auc_std.py.