def make_sim(seed,
beta,
calibration=True,
scenario=None,
delta_beta=1.6,
future_symp_test=None,
end_day=None,
verbose=0):
# Set the parameters
total_pop = 67.86e6 # UK population size
pop_size = 100e3 # Actual simulated population
pop_scale = int(total_pop / pop_size)
pop_type = 'hybrid'
pop_infected = 1500
beta = beta
asymp_factor = 2
contacts = {'h': 3.0, 's': 20, 'w': 20, 'c': 20}
beta_layer = {'h': 3.0, 's': 1.0, 'w': 0.6, 'c': 0.3}
if end_day is None: end_day = '2021-03-31'
pars = sc.objdict(
pop_size=pop_size,
pop_infected=pop_infected,
pop_scale=pop_scale,
pop_type=pop_type,
start_day=start_day,
end_day=end_day,
beta=beta,
asymp_factor=asymp_factor,
contacts=contacts,
rescale=True,
rand_seed=seed,
verbose=verbose,
rel_severe_prob=0.4,
rel_crit_prob=2.3,
#rel_death_prob=1.5,
)
sim = cv.Sim(pars=pars, datafile=data_path, location='uk')
sim['prognoses']['sus_ORs'][0] = 1.0 # ages 20-30
sim['prognoses']['sus_ORs'][1] = 1.0 # ages 20-30
# ADD BETA INTERVENTIONS
sbv = 0.63
beta_past = sc.odict({
'2020-02-14': [1.00, 1.00, 0.90, 0.90],
'2020-03-16': [1.00, 0.90, 0.80, 0.80],
'2020-03-23': [1.29, 0.02, 0.20, 0.20],
'2020-06-01': [1.00, 0.23, 0.40, 0.40],
'2020-06-15': [1.00, 0.38, 0.50, 0.50],
'2020-07-22': [1.29, 0.00, 0.30, 0.50],
'2020-09-02': [1.25, sbv, 0.50, 0.70],
'2020-10-01': [1.25, sbv, 0.50, 0.70],
'2020-10-16': [1.25, sbv, 0.50, 0.70],
'2020-10-26': [1.00, 0.00, 0.50, 0.70],
'2020-11-05': [1.25, sbv, 0.30, 0.40],
'2020-11-14': [1.25, sbv, 0.30, 0.40],
'2020-11-21': [1.25, sbv, 0.30, 0.40],
'2020-11-30': [1.25, sbv, 0.30, 0.40],
'2020-12-03': [1.50, sbv, 0.50, 0.70],
'2020-12-20': [1.25, 0.00, 0.50, 0.70],
'2020-12-25': [1.50, 0.00, 0.20, 0.90],
'2020-12-26': [1.50, 0.00, 0.20, 0.90],
'2020-12-31': [1.50, 0.00, 0.20, 0.90],
'2021-01-01': [1.50, 0.00, 0.20, 0.90],
'2021-01-04': [1.25, 0.14, 0.30, 0.40],
'2021-01-11': [1.25, 0.14, 0.30, 0.40],
'2021-01-18': [1.25, 0.14, 0.30, 0.40],
'2021-01-18': [1.25, 0.14, 0.30, 0.40]
})
if not calibration:
##no schools until 1st March but assue 20% (1 in 5) in schools between 04/01-22/02;
##model transmission remaining at schools as 14% (to account for 30% reduction due to school measures)
if scenario == 'FNL':
beta_s_feb22, beta_s_mar01, beta_s_mar08, beta_s_mar15, beta_s_mar22, beta_s_mar29, beta_s_apr01 = 0.14, 0.14, 0.14, 0.14, 0.14, 0.14, 0.02
##primaries and yars 11 and 13 back on 22/02 all other years 01/03
##9/14 years back -30% transmission reduction = 45% reduction remaining from 22/02
##transmision increases to 63% remaining from 01/03
##Easter holiday 01/04-08/04
elif scenario == 'staggeredPNL':
beta_s_feb22, beta_s_mar01, beta_s_mar08, beta_s_mar15, beta_s_mar22, beta_s_mar29, beta_s_apr01 = 0.14, 0.14, 0.40, sbv, sbv, sbv, 0.02,
##primaries and secondaries back fully 22/02; 14/14 years but assume 90% attendence and
##30% reduction in transmission due to hygiene, masks etc to remaining transmision to 0.63
##Easter holiday 01/04-08/04
elif scenario == 'fullPNL':
beta_s_feb22, beta_s_mar01, beta_s_mar08, beta_s_mar15, beta_s_mar22, beta_s_mar29, beta_s_apr01 = 0.14, 0.14, sbv, sbv, sbv, sbv, 0.02
elif scenario == 'primaryPNL':
beta_s_feb22, beta_s_mar01, beta_s_mar08, beta_s_mar15, beta_s_mar22, beta_s_mar29, beta_s_apr01 = 0.14, 0.14, 0.31, 0.31, 0.40, 0.40, 0.02
elif scenario == 'rotasecondaryPNL':
beta_s_feb22, beta_s_mar01, beta_s_mar08, beta_s_mar15, beta_s_mar22, beta_s_mar29, beta_s_apr01 = 0.14, 0.14, 0.31, 0.31, sbv, sbv, 0.02
beta_scens = sc.odict({
'2021-01-30': [1.25, 0.14, 0.30, 0.40],
'2021-02-08': [1.25, 0.14, 0.30, 0.40],
'2021-02-15': [1.25, 0.14, 0.30, 0.40],
'2021-02-22': [1.25, beta_s_feb22, 0.30, 0.40],
'2021-03-01': [1.25, beta_s_mar01, 0.30, 0.40],
'2021-03-08': [1.25, beta_s_mar08, 0.30, 0.50],
'2021-03-15': [1.25, beta_s_mar15, 0.30, 0.50],
'2021-03-22': [1.25, beta_s_mar22, 0.30, 0.50],
'2021-03-29': [1.25, beta_s_mar29, 0.30, 0.50],
'2021-04-01': [1.25, beta_s_apr01, 0.30, 0.50],
'2021-04-12': [1.25, 0.02, 0.30, 0.50],
'2021-04-19': [1.25, sbv, 0.50, 0.70],
'2021-04-26': [1.25, sbv, 0.50, 0.70],
'2021-05-03': [1.25, sbv, 0.50, 0.70]
})
beta_dict = sc.mergedicts(beta_past, beta_scens)
else:
beta_dict = beta_past
beta_days = list(beta_dict.keys())
h_beta = cv.change_beta(days=beta_days,
changes=[c[0] for c in beta_dict.values()],
layers='h')
s_beta = cv.change_beta(days=beta_days,
changes=[c[1] for c in beta_dict.values()],
layers='s')
w_beta = cv.change_beta(days=beta_days,
changes=[c[2] for c in beta_dict.values()],
layers='w')
c_beta = cv.change_beta(days=beta_days,
changes=[c[3] for c in beta_dict.values()],
layers='c')
# Add a new change in beta to represent the takeover of the novel variant VOC 202012/01
# Assume that the new variant is 60% more transmisible (https://cmmid.github.io/topics/covid19/uk-novel-variant.html,
# Assume that between Nov 1 and Jan 30, the new variant grows from 0-100% of cases
voc_days = np.linspace(sim.day('2020-08-01'), sim.day('2021-01-30'), 31)
voc_prop = 0.6 / (
1 + np.exp(-0.075 * (voc_days - sim.day('2020-09-30')))
) # Use a logistic growth function to approximate fig 2A of https://cmmid.github.io/topics/covid19/uk-novel-variant.html
voc_change = voc_prop * 1.63 + (1 - voc_prop) * 1.
voc_beta = cv.change_beta(days=voc_days, changes=voc_change)
interventions = [h_beta, w_beta, s_beta, c_beta, voc_beta]
# ADD TEST AND TRACE INTERVENTIONS
tc_day = sim.day(
'2020-03-16'
) #intervention of some testing (tc) starts on 16th March and we run until 1st April when it increases
te_day = sim.day(
'2020-04-01'
) #intervention of some testing (te) starts on 1st April and we run until 1st May when it increases
tt_day = sim.day(
'2020-05-01'
) #intervention of increased testing (tt) starts on 1st May
tti_day = sim.day(
'2020-06-01'
) #intervention of tracing and enhanced testing (tti) starts on 1st June
tti_day_july = sim.day(
'2020-07-01'
) #intervention of tracing and enhanced testing (tti) at different levels starts on 1st July
tti_day_august = sim.day(
'2020-08-01'
) #intervention of tracing and enhanced testing (tti) at different levels starts on 1st August
tti_day_sep = sim.day('2020-09-01')
tti_day_oct = sim.day('2020-10-01')
tti_day_nov = sim.day('2020-11-01')
tti_day_dec = sim.day('2020-12-01')
tti_day_jan = sim.day('2021-01-01')
tti_day_vac = sim.day('2020-12-20')
s_prob_april = 0.009
s_prob_may = 0.012
s_prob_june = 0.02769
s_prob_july = 0.02769
s_prob_august = 0.03769
tn = 0.09
s_prob_sept = 0.08769
s_prob_oct = 0.08769
s_prob_nov = 0.08769
s_prob_may = 0.02769
s_prob_june = 0.02769
s_prob_july = 0.02769
s_prob_august = 0.03769
s_prob_sep = 0.08769
s_prob_oct = 0.08769
s_prob_nov = 0.08769
s_prob_dec = 0.08769
if future_symp_test is None: future_symp_test = s_prob_dec
t_delay = 1.0
#isolation may-july
iso_vals = [{k: 0.1 for k in 'hswc'}]
#isolation august
iso_vals1 = [{k: 0.7 for k in 'hswc'}]
#isolation september
iso_vals2 = [{k: 0.5 for k in 'hswc'}]
#isolation october
iso_vals3 = [{k: 0.5 for k in 'hswc'}]
#isolation november
iso_vals4 = [{k: 0.5 for k in 'hswc'}]
#isolation december
iso_vals5 = [{k: 0.5 for k in 'hswc'}]
#testing and isolation intervention
interventions += [
cv.test_prob(symp_prob=0.0075,
asymp_prob=0.0,
symp_quar_prob=0.0,
start_day=tc_day,
end_day=te_day - 1,
test_delay=t_delay),
cv.test_prob(symp_prob=s_prob_april,
asymp_prob=0.0,
symp_quar_prob=0.0,
start_day=te_day,
end_day=tt_day - 1,
test_delay=t_delay),
cv.test_prob(symp_prob=s_prob_may,
asymp_prob=0.00076,
symp_quar_prob=0.0,
start_day=tt_day,
end_day=tti_day - 1,
test_delay=t_delay),
cv.test_prob(symp_prob=s_prob_june,
asymp_prob=0.00076,
symp_quar_prob=0.0,
start_day=tti_day,
end_day=tti_day_july - 1,
test_delay=t_delay),
cv.test_prob(symp_prob=s_prob_july,
asymp_prob=0.00076,
symp_quar_prob=0.0,
start_day=tti_day_july,
end_day=tti_day_august - 1,
test_delay=t_delay),
cv.test_prob(symp_prob=s_prob_august,
asymp_prob=0.0028,
symp_quar_prob=0.0,
start_day=tti_day_august,
end_day=tti_day_sep - 1,
test_delay=t_delay),
cv.test_prob(symp_prob=s_prob_sep,
asymp_prob=0.0028,
symp_quar_prob=0.0,
start_day=tti_day_sep,
end_day=tti_day_oct - 1,
test_delay=t_delay),
cv.test_prob(symp_prob=s_prob_oct,
asymp_prob=0.0028,
symp_quar_prob=0.0,
start_day=tti_day_oct,
end_day=tti_day_nov - 1,
test_delay=t_delay),
cv.test_prob(symp_prob=s_prob_nov,
asymp_prob=0.0063,
symp_quar_prob=0.0,
start_day=tti_day_nov,
end_day=tti_day_dec - 1,
test_delay=t_delay),
cv.test_prob(symp_prob=s_prob_dec,
asymp_prob=0.0063,
symp_quar_prob=0.0,
start_day=tti_day_dec,
end_day=tti_day_jan - 1,
test_delay=t_delay),
cv.test_prob(symp_prob=future_symp_test,
asymp_prob=0.0063,
symp_quar_prob=0.0,
start_day=tti_day_jan,
test_delay=t_delay),
cv.contact_tracing(trace_probs={
'h': 1,
's': 0.5,
'w': 0.5,
'c': 0.05
},
trace_time={
'h': 0,
's': 1,
'w': 1,
'c': 2
},
start_day='2020-06-01',
quar_period=10),
cv.dynamic_pars({'iso_factor': {
'days': te_day,
'vals': iso_vals
}}),
cv.dynamic_pars(
{'iso_factor': {
'days': tti_day_august,
'vals': iso_vals1
}}),
cv.dynamic_pars(
{'iso_factor': {
'days': tti_day_sep,
'vals': iso_vals2
}}),
cv.dynamic_pars(
{'iso_factor': {
'days': tti_day_oct,
'vals': iso_vals3
}}),
cv.dynamic_pars(
{'iso_factor': {
'days': tti_day_nov,
'vals': iso_vals4
}}),
cv.dynamic_pars(
{'iso_factor': {
'days': tti_day_dec,
'vals': iso_vals5
}})
]
#cv.dynamic_pars({'rel_death_prob': {'days': tti_day_vac, 'vals': 0.9}})]
#cv.vaccine(days=[0,14], rel_sus=0.4, rel_symp=0.2, cumulative=[0.7, 0.3])]
# vaccination interventions
interventions += [
utils.two_dose_daily_delayed(200e3,
start_day=tti_day_vac,
dose_delay=14,
delay=10 * 7,
take_prob=1.0,
rel_symp=0.05,
rel_trans=0.9,
cumulative=[0.7, 1.0],
dose_priority=[1, 0.1])
]
analyzers = []
analyzers += [
utils.record_dose_flows(vacc_class=utils.two_dose_daily_delayed)
]
# Finally, update the parameters
sim.update_pars(interventions=interventions, analyzers=analyzers)
# Change death and critical probabilities
# interventions += [cv.dynamic_pars({'rel_death_prob':{'days':sim.day('2020-07-01'), 'vals':0.6}})]
# Finally, update the parameters
#sim.update_pars(interventions=interventions)
for intervention in sim['interventions']:
intervention.do_plot = False
sim.initialize()
return sim
priority_days=[tti_day_vac1, tti_day_vac2, tti_day_vac3, tti_day_vac4, tti_day_vac5, tti_day_vac6], age_priority=[75, 60, 50, 40, 30, 20])]
# # vaccinating 50-65 years old
# interventions += [utils.two_dose_daily_delayed(300e3, start_day=tti_day_vac2, dose_delay=21, delay=12*7,
# take_prob=1.0, rel_symp=0.05,
# rel_trans=0.5, cumulative=[0.7, 1.0], dose_priority=[1, 0.1],
# priority_days=[tti_day_vac2, tti_day_vac3], age_priority=[50,40])]
#vaccinating 18-50 years old
#interventions += [utils.two_dose_daily_delayed(300e3, start_day=tti_day_vac3, dose_delay=21, delay=7*7,
# take_prob=1.0, rel_symp=0.05,
# rel_trans=0.5, cumulative=[0.7, 1.0], dose_priority=[1, 0.1],
# priority_days=[tti_day_vac3, tti_day_vac4], age_priority=[50,18])]
analyzers = []
analyzers += [utils.record_dose_flows(vacc_class=utils.two_dose_daily_delayed)]
analyzers += [cv.age_histogram(datafile='uk_stats_by_age.xlsx', edges=np.concatenate([np.linspace(0, 90, 19),np.array([100])]))]
# Finally, update the parameters
sim.update_pars(interventions=interventions, analyzers=analyzers)
# Change death and critical probabilities
# interventions += [cv.dynamic_pars({'rel_death_prob':{'days':sim.day('2020-07-01'), 'vals':0.6}})]
# Finally, update the parameters
#sim.update_pars(interventions=interventions)
for intervention in sim['interventions']: