Results from simulation of Dynamic Precision Medicine 2-Window Trial

Results from the simulation adapted from Beckman et al. PNAS 2012. For details see the preprint "A Generalized Evolutionary Classifier for Evolutionary Guided Precision Medicine"  https://www.medrxiv.org/content/10.1101/2020.09.24.20201111v2

The simualtion code that generated these results can be found at: https://github.com/GU-DPM/EvolutionaryClassifier

------------------------------------------------------------------

Annotations of simulation outcome files.

Simulation outcomes are stored in multiple files.

param_ALLDRUG_xxx.txt: parameter configuration values.stopt_ALLDRUG_xxx.txt: survival times for each parameter configuration and each strategy.dosage_ALLDRUG_xxx.txt: dosage combination sequences for each parameter configuration and each strategy.pop_ALLDRUG_xxx.txt: population composition dynamics for each parameter configuration and each strategy.

xxx denotes the index of the job when I ran simulations in my cluster.  You can simply concatenate the files over all indices.

param_ALLDRUG_xxx.txt format:

Each line denotes a parameter configuration with the following entries:

(1)parameter configuration index (one entry).(2)initial population compositions x0 (four entries, x0(S), x0(R1), x0(R2), x0(R12)).(3)growth rate g0 (one entry).(4)drug sensitivity matrix Sa (eight entries, Sa(S,drug1), Sa(S,drug2), Sa(R1,drug1), Sa(R1,drug2), ...).(5)transition rate matrix T (sixteen entries, T(S->S), T(S->R1), T(S->R2), T(S->R12), T(R1->S), T(R1->R1), T(R1->R2), T(R1->R12), ....

stopt_ALLDRUG_xxx.txt format:

Each line denotes the survival times (stopping times) of 10 strategies for each parameter configuration.

strategy 0: strategy 0 in the PNAS paper.strategy 1: strategy 2.2 in the PNAS paper.strategy 2: strategy 2.2 in the PNAS paper for first two treatment decisions, followed by strategy 0 in the PNAS paper

Each line has 11 entries: parameter configuration index and the survival times of the 10 strategies.

The time horizon of the therapy is 1800 days.  Each period is 45 days.  If the patient is cured (the total tumor population = 0), then the survival time is reported as 1845 days.  If the survival time is 1800 days, then the tumor population size is greater than 0 and smaller than the mortal level (1e13).

dosage_ALLDRUG_xxx.txt format:

Each line denotes the dosage combination sequence of each strategy for each parameter configuration.

(1)parameter configuration index.(2)strategy index (0-2).(3)(drug1 dosage,drug2 dosage) at t=0.(4)(drug1 dosage,drug2 dosage) at t=45....(42)(drug1 dosage,drug2 dosage) at t=1755.

If t exceeds the survival time of the strategy, then the drug dosage is set to -1.

pop_ALLDRUG_xxx.txt format:

Each line denotes the population composition dynamics of each strategy for each parameter configuration.

(1)parameter configuration index.(2)strategy index (0-9).(3)(S,R1,R1,R12) population size at t=45.(4)(S,R1,R1,R12) at t=90....(42)(S,R1,R1,R12) at t=1800.

If t exceeds the survival time of the strategy, then the population size is set to -1.

Notice the dosage combination is reported at the beginning of each period, and the population composition is reported at the end of each period.

When the best strategy over strategies 0-8 cures the patient and yields survival time 1800 days, I did not bother running strategy 9 (since it's time consuming).  Thus the dosage combination and population composition should be ignored.