HCC Detection, Reinforcement Learning Implementation

Replication study of 'Applying reinforcement learning techniques to detect hepatocellular carcinoma under limited screening capacity' with extension of MDP and DQN learning application.

Final report can be found here.

Reference Paper:

Lee E, Lavieri MS, Volk ML, and Xu Y. Applying reinforcement learning techniques to detect hepatocellular carcinoma under limited screening capacity. Health Care Manag Sci. 2015 Sep, 18(3):363-75. doi: 10.1007/s10729-014-9304-0. Epub 2014 Oct 12. PMID: 25308168.

Usage

Below is an example function call for the main python module:

Example 1: Single model simulation:

df = main(
        'data/subjects.csv', #points to the data file available
        T=3650, #planning horizon (days)
        D=90,   #decision epoch spacing (days)
        num_iters=20, #iterations to run per model simulation
        n=500, #panel size
        k=0.4, #constraint level
        policy='deep', #RL method name
        policy_params={}, #policy params if necessary (select policies only)
        replace=True, #sample with replacement to update panel
        d=0.05, #dropout rate of participants in study pool
        store_results=True #flag to store metrics in results/results.csv log
    )

Example 2: Run simulation for all models, model parameters, and constraint level permutations (this takes a long time!):

run_full_sim() #runs