Module uci_apc.multi_patient
Expand source code
import pandas as pd
import copy
import pathos.pools
import time
from simglucose.patient.t1dpatient import T1DPatient
from simglucose.simulation.sim_engine import SimObj
from simglucose.sensor.cgm import CGMSensor, CGMNoise
from simglucose.actuator.pump import InsulinPump
from simglucose.simulation.scenario_gen import RandomScenario
from simglucose.simulation.scenario import Action, CustomScenario
from simglucose.simulation.env import T1DSimEnv
from uci_apc.controller import PIDController
from datetime import timedelta, datetime
FRIENDLY_DATE_STR = str(datetime.strftime( datetime.now(), "%Y%m%d%H%M%S"))
def sim(sim_object):
'''
Simulate a sim object
Parameters
----------
sim_object: simglucose.simulation.sim_engine SimObj
Returns
-------
A pandas dataframe containing the simulation results.
axis=0: time, type datetime.datetime
axis=1: data category, type str
'''
print('simulating...')
sim_object.simulate()
return sim_object.results()
def run_sim_PID(no_runs, patients, runtime, meals, controller_params, path):
'''
Run the simulation a single time on a list of patients with the PID controller.
Parameters
----------
no_runs: int
the number of separate simulation runs.
patients: list of str
a list of patient name strings. Patient name strings can be found in the params/Quest.csv file inside simGlucose.
runtime: int
simulation time, in hours.
meals: (timedelta, int)
a tuple containing the time of meal (as referenced from simulation start) and the meal size, in grams.
targetBG: int
the target blood glucose for the controller, in mg/dl
lowBG: int
the pump suspension glucose for the controller, in mg/dl
Returns
-------
A pandas dataframe containing the simulation results.
axis=0: time, type datetime.datetime
axis=1: MultiIndex
level 0: data category, type str
level 1: patient id, type str
level 2: run number, type int (starts at 1)
'''
sensor = CGMSensor.withName('Dexcom')
pump = InsulinPump.withName('Insulet')
scenario = CustomScenario(start_time = datetime(2020, 1, 1, 0,0,0), scenario=meals)
sim_objs = []
keys = []
for run in range(0, no_runs):
for pname in patients:
sim_objs.append(SimObj(T1DSimEnv(T1DPatient.withName(pname),
sensor,
pump,
copy.deepcopy(scenario)), # because random numbers.
PIDController(controller_params, pname),
timedelta(hours=runtime),
animate=False,
path=None))
keys.append((run + 1, pname))
p_start = time.time()
print('Running batch simulation of {} items...'.format(len(patients * no_runs)))
p = pathos.pools.ProcessPool()
results = p.map(sim, sim_objs)
print('Simulation took {} seconds.'.format(time.time() - p_start))
return pd.concat(results, axis=1, keys=keys)
def run_sim_PID_once(pname, runtime, meals, controller_params, path):
'''
Run the simulation a single time on a single patient with the PID controller.
Parameters
----------
pname: str
patient name
runtime: int
simulation time, in hours.
meals: (timedelta, int)
a tuple containing the time of meal (as referenced from simulation start) and the meal size, in grams.
targetBG: int
the target blood glucose for the controller, in mg/dl
lowBG: int
the pump suspension glucose for the controller, in mg/dl
Returns
-------
A pandas dataframe containing the simulation results.
axis=0: time, type datetime.datetime
axis=1: data category, type str
'''
sensor = CGMSensor.withName('Dexcom')
pump = InsulinPump.withName('Insulet')
scenario = CustomScenario(start_time = datetime(2020, 1, 1, 0,0,0), scenario=meals)
obj = SimObj(T1DSimEnv(T1DPatient.withName(pname),
sensor,
pump,
scenario),
PIDController(controller_params, pname),
timedelta(hours=runtime),
animate=False,
path=None)
return sim(obj)
if __name__ == '__main__':
adolescents = ["adolescent#001","adolescent#002","adolescent#003","adolescent#004","adolescent#005","adolescent#006","adolescent#007","adolescent#008","adolescent#009","adolescent#010"]
children = ["child#001","child#002","child#003","child#004","child#005","child#006","child#007","child#008","child#009","child#010"]
adults = ["adult#001","adult#002","adult#003","adult#004","adult#005","adult#006","adult#007","adult#008","adult#009","adult#010"]
# (target, low, tau_c)
PIDparams = (120, 70, 100 )
t = 24
n = 40
meals = [(timedelta(hours=4), 80)]
pts = adults
dfs = run_sim_PID(n, adults, t, meals, PIDparams, './results/')
save=False
if save:
dfs_path = str('./results/' + FRIENDLY_DATE_STR + '-dfs.csv')
dfs.to_csv(dfs_path)
dfs.to_pickle('./results/' + 'adults_1-8_x40.bz2')
print(dfs)Functions
def run_sim_PID(no_runs, patients, runtime, meals, controller_params, path)-
Run the simulation a single time on a list of patients with the PID controller.
Parameters
no_runs:int- the number of separate simulation runs.
patients:listofstr- a list of patient name strings. Patient name strings can be found in the params/Quest.csv file inside simGlucose.
runtime:int- simulation time, in hours.
meals:(timedelta, int)- a tuple containing the time of meal (as referenced from simulation start) and the meal size, in grams.
targetBG:int- the target blood glucose for the controller, in mg/dl
lowBG:int- the pump suspension glucose for the controller, in mg/dl
Returns
A pandas dataframe containing the simulation results. axis=0: time, type datetime.datetime axis=1: MultiIndex level 0: data category, type str level 1: patient id, type str level 2: run number, type int (starts at 1)
Expand source code
def run_sim_PID(no_runs, patients, runtime, meals, controller_params, path): ''' Run the simulation a single time on a list of patients with the PID controller. Parameters ---------- no_runs: int the number of separate simulation runs. patients: list of str a list of patient name strings. Patient name strings can be found in the params/Quest.csv file inside simGlucose. runtime: int simulation time, in hours. meals: (timedelta, int) a tuple containing the time of meal (as referenced from simulation start) and the meal size, in grams. targetBG: int the target blood glucose for the controller, in mg/dl lowBG: int the pump suspension glucose for the controller, in mg/dl Returns ------- A pandas dataframe containing the simulation results. axis=0: time, type datetime.datetime axis=1: MultiIndex level 0: data category, type str level 1: patient id, type str level 2: run number, type int (starts at 1) ''' sensor = CGMSensor.withName('Dexcom') pump = InsulinPump.withName('Insulet') scenario = CustomScenario(start_time = datetime(2020, 1, 1, 0,0,0), scenario=meals) sim_objs = [] keys = [] for run in range(0, no_runs): for pname in patients: sim_objs.append(SimObj(T1DSimEnv(T1DPatient.withName(pname), sensor, pump, copy.deepcopy(scenario)), # because random numbers. PIDController(controller_params, pname), timedelta(hours=runtime), animate=False, path=None)) keys.append((run + 1, pname)) p_start = time.time() print('Running batch simulation of {} items...'.format(len(patients * no_runs))) p = pathos.pools.ProcessPool() results = p.map(sim, sim_objs) print('Simulation took {} seconds.'.format(time.time() - p_start)) return pd.concat(results, axis=1, keys=keys) def run_sim_PID_once(pname, runtime, meals, controller_params, path)-
Run the simulation a single time on a single patient with the PID controller.
Parameters
pname:str- patient name
runtime:int- simulation time, in hours.
meals:(timedelta, int)- a tuple containing the time of meal (as referenced from simulation start) and the meal size, in grams.
targetBG:int- the target blood glucose for the controller, in mg/dl
lowBG:int- the pump suspension glucose for the controller, in mg/dl
Returns
A pandas dataframe containing the simulation results. axis=0: time, type datetime.datetime axis=1: data category, type str
Expand source code
def run_sim_PID_once(pname, runtime, meals, controller_params, path): ''' Run the simulation a single time on a single patient with the PID controller. Parameters ---------- pname: str patient name runtime: int simulation time, in hours. meals: (timedelta, int) a tuple containing the time of meal (as referenced from simulation start) and the meal size, in grams. targetBG: int the target blood glucose for the controller, in mg/dl lowBG: int the pump suspension glucose for the controller, in mg/dl Returns ------- A pandas dataframe containing the simulation results. axis=0: time, type datetime.datetime axis=1: data category, type str ''' sensor = CGMSensor.withName('Dexcom') pump = InsulinPump.withName('Insulet') scenario = CustomScenario(start_time = datetime(2020, 1, 1, 0,0,0), scenario=meals) obj = SimObj(T1DSimEnv(T1DPatient.withName(pname), sensor, pump, scenario), PIDController(controller_params, pname), timedelta(hours=runtime), animate=False, path=None) return sim(obj) def sim(sim_object)-
Simulate a sim object
Parameters
sim_object:simglucose.simulation.sim_engine SimObj
Returns
A pandas dataframe containing the simulation results. axis=0: time, type datetime.datetime axis=1: data category, type str
Expand source code
def sim(sim_object): ''' Simulate a sim object Parameters ---------- sim_object: simglucose.simulation.sim_engine SimObj Returns ------- A pandas dataframe containing the simulation results. axis=0: time, type datetime.datetime axis=1: data category, type str ''' print('simulating...') sim_object.simulate() return sim_object.results()