-
Notifications
You must be signed in to change notification settings - Fork 0
/
03_Analysis_Japan_2groups_2assays.R
346 lines (273 loc) · 13.2 KB
/
03_Analysis_Japan_2groups_2assays.R
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
library(readstata13)
library(dplyr)
library(data.table)
library(ggplot2)
library(tibble)
library(knitr)
library(tidyr)
library(cowplot)
library(stringr)
library(randomForest)
library(RColorBrewer)
library(patchwork)
library(ggrepel)
library(rstan)
library(bayesplot)
library(readr)
library(lubridate)
library(brms)
library(incidental)
library(readxl)
library(GADMTools)
library(scales)
library(jsonlite)
library(tidyverse)
library(stringi)
source("Code/functions_for_gqs.R")
########################
## Load kinetics data ##
########################
## Load the bivariate results
FILENAME <- "Results/Stan_fit_bivariate_2groups.RData"
load(FILENAME)
## Look at traceplots
windows()
mcmc_trace(fit_Stan, pars=vars(contains("lambda"), contains("log_beta_severity"), contains("Sigma"), contains("tau"), contains("theta"))) + theme_bw()
## Look at sensitivity
windows()
as.data.frame(summary(fit_Stan)$summary) %>%
tibble::rownames_to_column() %>%
filter(str_detect(rowname, paste0("sensitivity_over_time\\["))) %>%
mutate(severity_final_class = as.numeric(str_match_all(rowname, "(?<=\\[).+?(?=\\,)"))) %>%
mutate(day = as.numeric(str_match_all(rowname, "(?<=\\,).+?(?=\\])"))) %>%
mutate(severity_final_class = factor(severity_final_class, levels=c(1:8), labels=c("Non-hosp (+,+)", "Non-hosp (+,-)", "Non-hosp (-,+)", "Non-hosp (-,-)",
"Hosp (+,+)", "Hosp (+,-)", "Hosp (-,+)", "Hosp (-,-)"))) %>%
ggplot() +
theme_bw() +
theme(legend.position="none") +
geom_ribbon(aes(x=day, ymin=`2.5%`, ymax=`97.5%`, group=severity_final_class, fill=severity_final_class), alpha=0.2) +
geom_line(aes(x=day, y=`50%`, group=severity_final_class, colour=severity_final_class), size=1) +
scale_x_continuous(breaks=scales::pretty_breaks(4)) +
scale_y_continuous(breaks=scales::pretty_breaks(5)) +
xlab("Days - 21") +
ylab("Sensitivity") +
ggtitle("(Abbott, Roche)") +
facet_wrap(~severity_final_class, scales="fixed", ncol=4)
## Fix parameters
prop_hosp <- 0.05683933
## Get the estimates of specificity
Sp_Abbott <- 0.9963
Sp_Roche <- 0.9980
##############################################################################################
## Calculate severity-weighted & epi curve-weighted sensitivity for each assay, and adjust: ##
## ##
## Japan: using EpiNow2 & deaths ##
##############################################################################################
## Read in merged EpiNow2 data
load("Results/EpiNow2_fit_Japan.RData")
backcalc_using_reported_deaths["summarised"] %>%
as.data.frame() %>%
filter(summarised.variable=="infections") %>%
droplevels() %>%
mutate(summarised.date = ymd(summarised.date)) %>%
rename(date = summarised.date) %>%
rename(onsets = summarised.median) %>%
select(date, onsets) %>%
droplevels() -> dat_ts_Japan
ex <- interval(ymd("2020-12-14"), ymd("2020-12-25"))
int_start(ex) + (int_end(ex) - int_start(ex))/2
## Remove 3 weeks
SURVEY_DATE_Japan <- ymd("2020-12-19")-weeks(3)
dat_ts_Japan %>%
filter(date <= SURVEY_DATE_Japan) %>%
mutate(days_before_survey = as.double(SURVEY_DATE_Japan - date)) %>%
mutate(prop_cases = onsets / sum(onsets)) -> dat_ts_Japan_final
## Bring in serosurvey results
dat_sero_Japan_bivariate <- read_excel("Data/Seroprevalence_data_Japan.xlsx", sheet=1)
## Get raw prevalence
dat_sero_Japan_bivariate$bivariate_raw <- dat_sero_Japan_bivariate$Positive_Both / dat_sero_Japan_bivariate$N_tested
dat_sero_Japan_bivariate$bivariate_median <- NA
dat_sero_Japan_bivariate$bivariate_lb <- NA
dat_sero_Japan_bivariate$bivariate_ub <- NA
## Read in model to generate new quantities (bivariate)
calc_wtd_Se_by_time_adjusting_for_severity_2groups_bivariate <-'
data {
real prop[2];
}
parameters {
matrix<lower=0.0, upper=1.0>[8,366] sensitivity_over_time;
}
generated quantities {
real sensitivity_wtd[4,366];
for(i in 1:366) {
sensitivity_wtd[1,i] = (prop[1]*sensitivity_over_time[1,i]) + (prop[2]*sensitivity_over_time[5,i]);
sensitivity_wtd[2,i] = (prop[1]*sensitivity_over_time[2,i]) + (prop[2]*sensitivity_over_time[6,i]);
sensitivity_wtd[3,i] = (prop[1]*sensitivity_over_time[3,i]) + (prop[2]*sensitivity_over_time[7,i]);
sensitivity_wtd[4,i] = (prop[1]*sensitivity_over_time[4,i]) + (prop[2]*sensitivity_over_time[8,i]);
}
}
'
model_weighted_Se <- stan_model(model_code=calc_wtd_Se_by_time_adjusting_for_severity_2groups_bivariate)
## Generate severity-weighted sensitivities
## This is ordered as (Abbott, Roche): (+,+), (+,-), (-,+), (-,-)
weighted_Se <- gqs(model_weighted_Se, data=list(prop=c(1-prop_hosp, prop_hosp)), draws=as.matrix(fit_Stan))
## Get the inputs: time series
N <- dat_ts_Japan_final %>% nrow()
N_times_4 <- N*4
prop_infections <- dat_ts_Japan_final %>% select(prop_cases) %>% pull()
## Get the inputs: sensitivity
## This is ordered as (Abbott, Roche): (+,+), (+,-), (-,+), (-,-)
dat_Se_Japan <- as.data.frame(weighted_Se) %>% select(contains("wtd")) %>% select(1:N_times_4) %>% as.matrix()
## Read in model to generate new quantities, for the bivariate situation
calc_wtd_Se_by_time_adjusting_for_severity_and_ts_bivariate <-'
data {
int N;
real prop_infections[N];
}
parameters {
real sensitivity_wtd[4,N];
}
generated quantities {
real sensitivity_wtd_with_ts[4];
sensitivity_wtd_with_ts[1] = 0.0;
sensitivity_wtd_with_ts[2] = 0.0;
sensitivity_wtd_with_ts[3] = 0.0;
sensitivity_wtd_with_ts[4] = 0.0;
for(i in 1:N) {
sensitivity_wtd_with_ts[1] += sensitivity_wtd[1,i] * prop_infections[N+1-i];
sensitivity_wtd_with_ts[2] += sensitivity_wtd[2,i] * prop_infections[N+1-i];
sensitivity_wtd_with_ts[3] += sensitivity_wtd[3,i] * prop_infections[N+1-i];
sensitivity_wtd_with_ts[4] += sensitivity_wtd[4,i] * prop_infections[N+1-i];
}
}
'
model_weighted_Se_with_ts <- stan_model(model_code=calc_wtd_Se_by_time_adjusting_for_severity_and_ts_bivariate)
## Estimate the parameters of Se (mean and sd), by assay
weighted_Se <- gqs(model_weighted_Se_with_ts, draws=dat_Se_Japan, data=list(N=N, prop_infections=prop_infections))
## Get the mu and sigma for each Se
mu_Se_1 <- summary(weighted_Se)$summary["sensitivity_wtd_with_ts[1]","mean"]
mu_Se_2 <- summary(weighted_Se)$summary["sensitivity_wtd_with_ts[2]","mean"]
mu_Se_3 <- summary(weighted_Se)$summary["sensitivity_wtd_with_ts[3]","mean"]
mu_Se_4 <- summary(weighted_Se)$summary["sensitivity_wtd_with_ts[4]","mean"]
sigma_Se_1 <- summary(weighted_Se)$summary["sensitivity_wtd_with_ts[1]","sd"]
sigma_Se_2 <- summary(weighted_Se)$summary["sensitivity_wtd_with_ts[2]","sd"]
sigma_Se_3 <- summary(weighted_Se)$summary["sensitivity_wtd_with_ts[3]","sd"]
sigma_Se_4 <- summary(weighted_Se)$summary["sensitivity_wtd_with_ts[4]","sd"]
## Model adapted from: https://pubmed.ncbi.nlm.nih.gov/10802336/
Stan_model_cond_covariance <- '
data {
// This is ordered as (Abbott, Roche): (+,+), (+,-), (-,+), (-,-)
real mu_Se[4];
real sigma_Se[4];
//real Se[4];
// This is ordered as (Abbott, Roche): (+,+), (+,-), (-,+), (-,-)
int y_sample_Tokyo[4];
int y_sample_Osaka[4];
int y_sample_Miyagi[4];
int y_sample_Aichi[4];
int y_sample_Fukuoka[4];
real Sp_Abbott;
real Sp_Roche;
}
parameters {
real<lower=0.0, upper=1.0> prev_Tokyo;
real<lower=0.0, upper=1.0> prev_Osaka;
real<lower=0.0, upper=1.0> prev_Miyagi;
real<lower=0.0, upper=1.0> prev_Aichi;
real<lower=0.0, upper=1.0> prev_Fukuoka;
simplex[4] Se;
}
transformed parameters {
simplex[4] p_sample_Tokyo;
simplex[4] p_sample_Osaka;
simplex[4] p_sample_Miyagi;
simplex[4] p_sample_Aichi;
simplex[4] p_sample_Fukuoka;
p_sample_Tokyo[1] = prev_Tokyo*(Se[1]) + (1.0-prev_Tokyo)*(1.0-Sp_Abbott)*(1.0-Sp_Roche);
p_sample_Tokyo[2] = prev_Tokyo*(Se[2]) + (1.0-prev_Tokyo)*(1.0-Sp_Abbott)*Sp_Roche;
p_sample_Tokyo[3] = prev_Tokyo*(Se[3]) + (1.0-prev_Tokyo)*Sp_Abbott*(1.0-Sp_Roche);
p_sample_Tokyo[4] = prev_Tokyo*(Se[4]) + (1.0-prev_Tokyo)*Sp_Abbott*Sp_Roche;
p_sample_Osaka[1] = prev_Osaka*(Se[1]) + (1.0-prev_Osaka)*(1.0-Sp_Abbott)*(1.0-Sp_Roche);
p_sample_Osaka[2] = prev_Osaka*(Se[2]) + (1.0-prev_Osaka)*(1.0-Sp_Abbott)*Sp_Roche;
p_sample_Osaka[3] = prev_Osaka*(Se[3]) + (1.0-prev_Osaka)*Sp_Abbott*(1.0-Sp_Roche);
p_sample_Osaka[4] = prev_Osaka*(Se[4]) + (1.0-prev_Osaka)*Sp_Abbott*Sp_Roche;
p_sample_Miyagi[1] = prev_Miyagi*(Se[1]) + (1.0-prev_Miyagi)*(1.0-Sp_Abbott)*(1.0-Sp_Roche);
p_sample_Miyagi[2] = prev_Miyagi*(Se[2]) + (1.0-prev_Miyagi)*(1.0-Sp_Abbott)*Sp_Roche;
p_sample_Miyagi[3] = prev_Miyagi*(Se[3]) + (1.0-prev_Miyagi)*Sp_Abbott*(1.0-Sp_Roche);
p_sample_Miyagi[4] = prev_Miyagi*(Se[4]) + (1.0-prev_Miyagi)*Sp_Abbott*Sp_Roche;
p_sample_Aichi[1] = prev_Aichi*(Se[1]) + (1.0-prev_Aichi)*(1.0-Sp_Abbott)*(1.0-Sp_Roche);
p_sample_Aichi[2] = prev_Aichi*(Se[2]) + (1.0-prev_Aichi)*(1.0-Sp_Abbott)*Sp_Roche;
p_sample_Aichi[3] = prev_Aichi*(Se[3]) + (1.0-prev_Aichi)*Sp_Abbott*(1.0-Sp_Roche);
p_sample_Aichi[4] = prev_Aichi*(Se[4]) + (1.0-prev_Aichi)*Sp_Abbott*Sp_Roche;
p_sample_Fukuoka[1] = prev_Fukuoka*(Se[1]) + (1.0-prev_Fukuoka)*(1.0-Sp_Abbott)*(1.0-Sp_Roche);
p_sample_Fukuoka[2] = prev_Fukuoka*(Se[2]) + (1.0-prev_Fukuoka)*(1.0-Sp_Abbott)*Sp_Roche;
p_sample_Fukuoka[3] = prev_Fukuoka*(Se[3]) + (1.0-prev_Fukuoka)*Sp_Abbott*(1.0-Sp_Roche);
p_sample_Fukuoka[4] = prev_Fukuoka*(Se[4]) + (1.0-prev_Fukuoka)*Sp_Abbott*Sp_Roche;
}
model {
// Sero-prevalence
target += multinomial_lpmf(y_sample_Tokyo | p_sample_Tokyo);
target += multinomial_lpmf(y_sample_Osaka | p_sample_Osaka);
target += multinomial_lpmf(y_sample_Miyagi | p_sample_Miyagi);
target += multinomial_lpmf(y_sample_Aichi | p_sample_Aichi);
target += multinomial_lpmf(y_sample_Fukuoka | p_sample_Fukuoka);
for(i in 1:4) target += normal_lpdf(Se[i] | mu_Se[i], sigma_Se[i]);
}
'
## Fit the model
fit_Stan_cond_covariance <- stan(
model_code=Stan_model_cond_covariance,
data=list(
## This is ordered as (Abbott, Roche): (+,+), (+,-), (-,+), (-,-)
mu_Se = c(mu_Se_1, mu_Se_2, mu_Se_3, mu_Se_4),
sigma_Se = c(sigma_Se_1, sigma_Se_2, sigma_Se_3, sigma_Se_4),
## This is ordered as (Abbott, Roche): (+,+), (+,-), (-,+), (-,-)
y_sample_Tokyo=c(31,6,29,3333),
y_sample_Osaka=c(16,5,9,2716),
y_sample_Miyagi=c(4,8,5,2843),
y_sample_Aichi=c(16,9,11,2924),
y_sample_Fukuoka=c(6,6,13,3053),
## From package inserts
Sp_Abbott=Sp_Abbott,
Sp_Roche=Sp_Roche),
chains=4, iter=5000, thin=5, init=0, control=list(adapt_delta=0.9))
## Store
dat_sero_Japan_bivariate$bivariate_median[1] <- summary(fit_Stan_cond_covariance)$summary["prev_Tokyo","50%"]
dat_sero_Japan_bivariate$bivariate_lb[1] <- summary(fit_Stan_cond_covariance)$summary["prev_Tokyo","2.5%"]
dat_sero_Japan_bivariate$bivariate_ub[1] <- summary(fit_Stan_cond_covariance)$summary["prev_Tokyo","97.5%"]
dat_sero_Japan_bivariate$bivariate_median[2] <- summary(fit_Stan_cond_covariance)$summary["prev_Osaka","50%"]
dat_sero_Japan_bivariate$bivariate_lb[2] <- summary(fit_Stan_cond_covariance)$summary["prev_Osaka","2.5%"]
dat_sero_Japan_bivariate$bivariate_ub[2] <- summary(fit_Stan_cond_covariance)$summary["prev_Osaka","97.5%"]
dat_sero_Japan_bivariate$bivariate_median[3] <- summary(fit_Stan_cond_covariance)$summary["prev_Miyagi","50%"]
dat_sero_Japan_bivariate$bivariate_lb[3] <- summary(fit_Stan_cond_covariance)$summary["prev_Miyagi","2.5%"]
dat_sero_Japan_bivariate$bivariate_ub[3] <- summary(fit_Stan_cond_covariance)$summary["prev_Miyagi","97.5%"]
dat_sero_Japan_bivariate$bivariate_median[4] <- summary(fit_Stan_cond_covariance)$summary["prev_Aichi","50%"]
dat_sero_Japan_bivariate$bivariate_lb[4] <- summary(fit_Stan_cond_covariance)$summary["prev_Aichi","2.5%"]
dat_sero_Japan_bivariate$bivariate_ub[4] <- summary(fit_Stan_cond_covariance)$summary["prev_Aichi","97.5%"]
dat_sero_Japan_bivariate$bivariate_median[5] <- summary(fit_Stan_cond_covariance)$summary["prev_Fukuoka","50%"]
dat_sero_Japan_bivariate$bivariate_lb[5] <- summary(fit_Stan_cond_covariance)$summary["prev_Fukuoka","2.5%"]
dat_sero_Japan_bivariate$bivariate_ub[5] <- summary(fit_Stan_cond_covariance)$summary["prev_Fukuoka","97.5%"]
## Plot
windows()
dat_sero_Japan_bivariate %>%
pivot_longer(cols=bivariate_raw:bivariate_ub) %>%
separate(name, sep="_", into=c("Assay","metric")) %>%
filter(metric=="raw") %>%
mutate(metric = "raw") %>%
mutate(Assay = "Bivariate") %>%
ggplot() +
geom_pointrange(data=dat_sero_Japan_bivariate %>% mutate(Assay="Bivariate") %>% mutate(metric="adjusted"), aes(x=Assay, y=bivariate_median, ymin=bivariate_lb, ymax=bivariate_ub, colour=metric)) +
# geom_point(data=dat_sero_Japan_bivariate %>% mutate(Assay="bivariate") %>% mutate(metric="raw (Abbott +)"), aes(x=Assay, y=Positive_Abbott/N_tested, colour=metric), size=2.5) +
# geom_point(data=dat_sero_Japan_bivariate %>% mutate(Assay="bivariate") %>% mutate(metric="raw (Roche +)"), aes(x=Assay, y=Positive_Roche/N_tested, colour=metric), size=2.5) +
geom_point(aes(x=Assay, y=value, colour=metric), size=2.5) +
scale_colour_manual(values=c("raw (Abbott +)"="black", "raw (Roche +)"="black", "raw"="black", "adjusted"="red"), breaks=c("adjusted","raw (Roche +)","raw (Abbott +)","raw")) +
theme_bw() +
xlab("") +
ylab("Seroprevalence") +
facet_wrap(.~Prefecture, scales="fixed", ncol=5)
## Prep to save
dat_sero_Japan_bivariate_2groups <- dat_sero_Japan_bivariate
# Save for future plotting etc
save(
dat_sero_Japan_bivariate_2groups,
file="Results/Results_Japan_2groups_2assays.RData")