def __init__(self, csv_file_name, wtps, poly_degree=2):
scenario_df = Cls.ScenarioDataFrame(csv_file_name=csv_file_name)
self.wtps = wtps
self.selectWTPs = []
self.costs = []
self.effects = []
self.utilization = []
self.nSwitches = []
for scenario_name in scenario_df.scenarios:
if scenario_name[0:3] == 'D:2':
# store wtp value
i = scenario_name.find('WTP')
self.selectWTPs.append(float(scenario_name[i + 5:]))
# total cost
cost_mean, cost_CI = scenario_df.get_mean_interval(
scenario_name=scenario_name, outcome_name='Total Cost')
self.costs.append(cost_mean)
# QALY
effect_mean, effect_CI = scenario_df.get_mean_interval(
scenario_name=scenario_name, outcome_name='DALY')
self.effects.append(effect_mean)
# utilization of social distancing
utilization_mean, utilization_CI = scenario_df.get_mean_interval(
scenario_name=scenario_name,
outcome_name='Utilization (unit of time): Social Distancing'
)
self.utilization.append(utilization_mean)
# Number of Switches
n_switches_mean, n_switches_CI = scenario_df.get_mean_interval(
scenario_name=scenario_name,
outcome_name='Number of Switches')
self.nSwitches.append(n_switches_mean)
self.costRegression = Reg.PolyRegression(
self.selectWTPs, [c * 1e-6 for c in self.costs],
degree=poly_degree)
self.dalyRegression = Reg.PolyRegression(
self.selectWTPs, [e * 1e-3 for e in self.effects],
degree=poly_degree)
self.utilRegression = Reg.PolyRegression(self.selectWTPs,
self.utilization,
degree=poly_degree)
self.nSwitchesRegression = Reg.PolyRegression(self.selectWTPs,
self.nSwitches,
degree=1)
def __init__(self, csv_file_name, poly_degree=2):
scenario_df = Cls.ScenarioDataFrame(csv_file_name=csv_file_name)
self.wtps = []
self.costs = []
self.effects = []
self.nSwitches = []
for scenario_name in scenario_df.scenarios:
if scenario_name[0:3] == 'D:3':
# store wtp value
i = scenario_name.find('WTP')
self.wtps.append(float(scenario_name[i + 5:]))
# total cost
cost_mean, cost_CI = scenario_df.get_mean_interval(
scenario_name=scenario_name, outcome_name='Total Cost')
self.costs.append(cost_mean)
# effect
effect_mean, effect_CI = scenario_df.get_mean_interval(
scenario_name=scenario_name, outcome_name='DALY')
self.effects.append(effect_mean / 10)
# Number of Switches
n_switches_mean, n_switches_CI = scenario_df.get_mean_interval(
scenario_name=scenario_name,
outcome_name='Number of Switches')
self.nSwitches.append(n_switches_mean)
self.costRegression = Reg.PolyRegression(self.wtps,
self.costs,
degree=poly_degree)
self.costRegression = Reg.ExpRegression(self.wtps, self.costs)
self.effectRegression = Reg.PolyRegression(self.wtps,
self.effects,
degree=poly_degree)
self.effectRegression = Reg.ExpRegression(
self.wtps, self.effects, if_c0_zero=True) # p0=[0, 1, -1])
self.nSwitchesRegression = Reg.PolyRegression(self.wtps,
self.nSwitches,
degree=poly_degree)
import apace.ScenariosClasses as Cls
import apace.VisualizeScenarios as Vis
import covid19.Support as Sup
import SimPy.FormatFunctions as F
SUFX = ''
# SUFX = '4WeekMin'
Cls.POLY_DEGREES = 2
scenarioDfFixedPeriodic = Cls.ScenarioDataFrame(
csv_file_name='csv_files/PolicyEvals/PolicyEvalsFixedPeriodic{}.csv'.
format(SUFX))
scenarioDfAdaptiveI = Cls.ScenarioDataFrame(
csv_file_name='csv_files/PolicyEvals/PolicyEvalAdaptiveFt{}.csv'.format(
SUFX))
policy_definitions = Sup.PolicyDefinitions()
# series to display on the cost-effectiveness plane
fixed_interval = Cls.SetOfScenarios(
name='Predetermined Duration',
scenario_df=scenarioDfFixedPeriodic,
color='blue',
marker='o',
conditions_on_variables=policy_definitions.FixedIntervalVarConditions,
if_find_frontier=False,
if_show_fitted_curve=True,
labels_shift_x=-0.025,
labels_shift_y=0.02)
periodic = Cls.SetOfScenarios(
import apace.ScenariosClasses as Cls
import apace.VisualizeScenarios as Vis
import covid19.Support as Sup
import SimPy.FormatFunctions as F
SUFX = ''
# SUFX = '4WeekMin'
Cls.POLY_DEGREES = 2
scenarioDfFixedPeriodic = Cls.ScenarioDataFrame(
csv_file_name='csv_files/PolicyEvals/PolicyEvalsFixedPeriodic{}.csv'.format(SUFX))
scenarioDfAdaptiveIt = Cls.ScenarioDataFrame(
csv_file_name='csv_files/PolicyEvals/PolicyEvalAdaptiveIt{}.csv'.format(SUFX))
policy_definitions = Sup.PolicyDefinitions()
# series to display on the cost-effectiveness plane
fixed_interval = Cls.SetOfScenarios(name='Static: Continuous',
scenario_df=scenarioDfFixedPeriodic,
color='blue',
marker='o',
conditions_on_variables=policy_definitions.FixedIntervalVarConditions,
if_find_frontier=False,
if_show_fitted_curve=True,
labels_shift_x=0.01,
labels_shift_y=0.01)
periodic = Cls.SetOfScenarios(name='Static: Periodic',
scenario_df=scenarioDfFixedPeriodic,
color='purple',
import SimPy.SensitivityAnalysis as SA
import SimPy.InOutFunctions as IO
from apace import ScenariosClasses as Sce
scenario_keys = [
'Base', '75% PTFU | No >1 FU | Drop % | No IPT',
'75% PTFU | With >1 FU | Drop 15% | No IPT',
'75% PTFU | No >1 FU | Drop % | With IPT',
'75% PTFU | With >1 FU | Drop 15% | With IPT'
]
# data frame for scenario analysis
scenario_df = Sce.ScenarioDataFrame('csv_files\TBScenarios.csv')
# read parameter samples
parameter_values = IO.read_csv_cols_to_dictionary(
file_name='csv_files/SampledParams.csv',
delimiter=',',
if_convert_float=True)
# create a dictionaries of DALYs and cost
dict_DALY = {}
dict_cost = {}
for key in scenario_keys:
dict_DALY[key] = scenario_df.scenarios[key].outcomes['DALY']
dict_cost[key] = scenario_df.scenarios[key].outcomes['Total Cost']
# create dictionaries for dDALYS and dCost
dict_dDALY = {}
dict_dCost = {}
for key in scenario_keys:
def __init__(self, if_spatial=False):
# conditions of variables to define scenarios to display
# on each series of cost-effectiveness plane
self.FixedIntervalVarConditions = [
Cls.ConditionOnVariable('Decision Rule',
0,
0,
if_included_in_label=False),
Cls.ConditionOnVariable(
'Duration of Social Distancing',
0,
110, # 200
if_included_in_label=True,
label_format='{:.0f}'),
# Cls.VariableCondition(' Time of lifting social distancing', 1, 1,
# if_included_in_label=False),
# Cls.VariableCondition('Switch threshold if social distancing is off',
# DS_TESTS, DS_TESTS,
# if_included_in_label=False),
# Cls.VariableCondition('Switch threshold if social distancing is on',
# DS_TESTS, DS_TESTS,
# if_included_in_label=False)
]
self.ICUInfVarConditions = [
Cls.ConditionOnVariable('Decision Rule',
2,
2,
if_included_in_label=False),
Cls.ConditionOnVariable(
'% I Switch threshold if social distancing is off',
0,
20000, # 1, 5
if_included_in_label=False,
label_format='{:.0f}'),
Cls.ConditionOnVariable(
'% I Switch threshold if social distancing is on',
0,
20000, # 0, 5
if_included_in_label=False,
label_format='{:.0f}'),
Cls.ConditionOnVariable(
'WTP',
0,
0, # 0, 5
if_included_in_label=False,
label_format='{:.0f}'),
# Cls.VariableCondition(' Time of lifting social distancing', 1, 1,
# if_included_in_label=False),
# Cls.VariableCondition('Switch threshold if social distancing is off',
# DS_TESTS, DS_TESTS,
# if_included_in_label=False),
# Cls.VariableCondition('Switch threshold if social distancing is on',
# DS_TESTS, DS_TESTS,
# if_included_in_label=False)
]
self.ICUInfOutcomeConditions = [
Cls.ConditionOnOutcome(
outcome_name='Average ratio: % served in ICU',
minimum=0.4,
maximum=1,
if_included_in_label=True,
label_format='{:.0%}'),
Cls.ConditionOnOutcome(outcome_name='Number of Switches',
minimum=0,
maximum=100,
if_included_in_label=False,
label_format='{:.0f}')
]
self.PeriodicVarConditions = [
Cls.ConditionOnVariable('Decision Rule',
1,
1,
if_included_in_label=False),
Cls.ConditionOnVariable(
'Periodicity (weeks)', #2, 8, # 1, 5
values=[2, 4, 8],
if_included_in_label=True,
label_format='{:.0f}'),
# Cls.VariableCondition(' Time of lifting social distancing', 1, 1,
# if_included_in_label=False),
# Cls.VariableCondition('Switch threshold if social distancing is off',
# DS_TESTS, DS_TESTS,
# if_included_in_label=False),
# Cls.VariableCondition('Switch threshold if social distancing is on',
# DS_TESTS, DS_TESTS,
# if_included_in_label=False)
]
self.AdaptiveItVarConditions = [
Cls.ConditionOnVariable('Decision Rule',
3,
3,
if_included_in_label=False),
Cls.ConditionOnVariable(
'WTP', #0.01, 2, # 1, 5
values=[
0.01, 0.02, 0.03, 0.04, 0.05, 0.06, 0.08, 0.1, 0.12, 0.13,
0.14
],
if_included_in_label=True,
label_format='{:.2f}'),
# Cls.VariableCondition(' Time of lifting social distancing', 1, 1,
# if_included_in_label=False),
# Cls.VariableCondition('Switch threshold if social distancing is off',
# DS_TESTS, DS_TESTS,
# if_included_in_label=False),
# Cls.VariableCondition('Switch threshold if social distancing is on',
# DS_TESTS, DS_TESTS,
# if_included_in_label=False)
]
self.AdaptiveRVarConditions = [
Cls.ConditionOnVariable('Decision Rule',
2,
2,
if_included_in_label=False),
Cls.ConditionOnVariable(
'R_t Switch threshold if social distancing is off',
1,
5,
if_included_in_label=False,
label_format='{:.1f}'),
Cls.ConditionOnVariable(
'R_t Switch threshold if social distancing is on',
0,
5,
if_included_in_label=False,
label_format='{:.1f}'),
# Cls.VariableCondition(' Time of lifting social distancing', 1, 1,
# if_included_in_label=False),
# Cls.VariableCondition('Switch threshold if social distancing is off',
# DS_TESTS, DS_TESTS,
# if_included_in_label=False),
# Cls.VariableCondition('Switch threshold if social distancing is on',
# DS_TESTS, DS_TESTS,
# if_included_in_label=False)
]
def analyze_econ_eval(prob_uptake, prob_dropout):
# find the name of this analysis
analysis_name = 'U{:.{prec}f}% '.format(prob_uptake * 100, prec=0) \
+ 'D{:.{prec}f}%'.format(prob_dropout * 100, prec=0)
print('Results for:', analysis_name)
# conditions of variables to define scenarios to display on the cost-effectiveness plane
# here we want scenarios with
# 'Prob {Tc+ | Tc}' = PROB,
# 'Follow-Up (Tc+>1)' = any, and
# 'IPT' = any
varConditions = [
Cls.ConditionOnVariable('Prob {Tc+ | Tc}',
minimum=prob_uptake,
maximum=prob_uptake,
if_included_in_label=False),
Cls.ConditionOnVariable('Follow-Up (Tc+>1)',
minimum=0,
maximum=1,
if_included_in_label=True,
label_rules=[
(0, 'First-year follow-up'),
(1, 'Annual follow-up')]
),
Cls.ConditionOnVariable('Prob {Drop-Out in Tc+>1}',
values=(prob_dropout, 0),
if_included_in_label=False),
Cls.ConditionOnVariable('IPT',
minimum=0,
maximum=1,
if_included_in_label=True,
label_rules=[
(0, ''),
(1, 'with 2°IPT')]
)
]
# data frame of scenarios
dfScenarios = Cls.ScenarioDataFrame(csv_file_name='csv_files\TBScenarios.csv')
# series to display on the cost-effectiveness plane
series = [
Cls.SetOfScenarios(name='U{:.{prec}f}%'.format(prob_uptake * 100, prec=0)
+'-D{:.{prec}f}%'.format(prob_dropout * 100, prec=0),
scenario_df=dfScenarios,
color='#4D4D4D', # '#808A87',
conditions_on_variables=varConditions,
if_find_frontier=True,
labels_shift_x=-0.04,
labels_shift_y=0.01)
]
# populate series
Cls.SetOfScenarios.populate_sets_of_scenarios(
series,
save_cea_results=True,
colors_of_scenarios=colors,
interval_type='p',
effect_multiplier=1,
cost_multiplier=1 / 1e3,
wtp_range=[0, 1000])
# CBA
#del series[0].CBA.strategies[1:3]
plt.rc('font', size=9) # fontsize of texts
series[0].CBA.plot_incremental_nmbs(
title='',
y_label='Incremental Net Monetary Benefit (Million $)',
x_label='Cost-Effectiveness Threshold ($ per DALY Averted)',
interval_type='n',
transparency_lines=0.5,
show_legend=True,
y_range=(-1, 5),
y_axis_multiplier=1/1000000, y_axis_decimal=1,
figure_size=(4, 3.6),
file_name='results/cea/NMB-' + analysis_name + '.pdf'
)
print('WTP range with the highest expected NMB:')
print(series[0].CBA.find_optimal_switching_wtp_values(deci=1))
series[0].CBA.plot_acceptability_curves(
#title='Cost-Effectiveness Acceptability Curves',
x_label='Cost-Effectiveness Threshold ($ per DALY Averted)',
y_label='Probability of Resulting in the Highest NMB',
y_range=None,
fig_size=(4, 3.6),
legends=['Base',
'First-year follow-up',
'Annual follow-up',
'First-year follow-up with limited 2°IPT',
'Annual follow-up with continuous 2°IPT'],
file_name='results/cea/CEAC ' + analysis_name + '.pdf'
)
print('WTP range with the highest probability of being optimal:')
# print(series[0].CBA.get_wtp_range_with_highest_prob_of_optimal())
# print dCost, dEffect and cost-effectiveness ratio with respect to the base
print('\nRelative cost to Base, Relative DALY to base, CER')
print(series[0].CEA.get_dCost_dEffect_cer(interval_type='p',
alpha=0.05,
cost_digits=0, effect_digits=0, icer_digits=1,
cost_multiplier=1, effect_multiplier=1))
# plot both incremental NMB and CEAC
plot_nmb_and_ceac(cba=series[0].CBA,
fig_size=(8, 3.5),
file_name='results\cea\INMB_CEAC-' + analysis_name + '.pdf')
# plot cost-effectiveness figure
plot_ce_figure(series=series, analysis_name=analysis_name)
# pairwise
# column titles
titles = ['Base',
'First-year follow-up',
'Annual follow-up',
'First-year follow-up\nwith limited 2°IPT',
'Annual follow-up\nwith continuous 2°IPT']
# plot
series[0].CEA.plot_pairwise_ceas(
figure_size=(7, 7),
font_size=7,
effect_label='DALY Averted (Thousands)',
cost_label='Additional Cost (Thousand Dollars)',
center_s=40,
cloud_s=10,
transparency=0.1,
effect_multiplier=1/1000,
cost_multiplier=1/1000,
x_range=[-10, 20],
y_range=[-2000, 2000],
column_titles=titles,
row_titles=titles,
file_name='results\cea\pairwise-' + analysis_name + '.pdf'
)