-
Notifications
You must be signed in to change notification settings - Fork 42
/
.Rhistory
512 lines (512 loc) · 26.4 KB
/
.Rhistory
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
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
t1 <- Sys.time() - p1 # calculate time to run the analusis by extracting p from the current system time
################ probabilistic analysis method 0.1 ###########################################
####### Generate random values inside a Markov function - use sapply #################################
p2 <- Sys.time() # save system time
set.seed(1) # set a seed to be able to reproduce the same results
Markov <- function(i) { # Markov model function - for different values of index i calculate the Markov model
p.HD <- rbeta(1, 20, 980) # probability from healthy to dead
p.HS <- rbeta(1, 50, 950) # probability from healthy to sick
p.SD <- rbeta(1, 100,900) # probability from sick to dead
c.H <- rnorm(1, 400, 50) # cost of being in healthy state
c.S <- rnorm(1, 100, 80) # cost of being in sick state
c.D <- 0 # cost of being dead
u.H <- rnorm(1, 0.8, 0.02) # utility of being in healthy state
u.S <- rnorm(1, 0.5, 0.02) # utility of being in the sick state
u.D <- 0 # utility of being dead
m.P <- matrix(c(1 - p.HD - p.HS, p.HS, p.HD, # transition matrix
0, 1 - p.SD, p.SD,
0, 0, 1),
byrow = TRUE, nrow = n.s,
dimnames = list(v.n, v.n))
m.TR <- matrix(NA, nrow = n.t + 1, ncol = n.s,
dimnames = list(0:n.t, v.n)) # create Markov trace
m.TR[1, ] <- c(1, 0, 0) # initialize Markov trace
for (t in 1:n.t) { # throughout the number of cycles
m.TR[t + 1, ] <- m.TR[t, ] %*% m.P # estimate the Markov trace for cycle t + 1
} # close the for loop for time
v.c <- c(c.H, c.S, c.D) # vector of costs
v.e <- c(u.H, u.S, u.D) # vector of utilities
TC <- t(m.TR %*% v.c) %*% v.dw # total cost for ith simulation
TE <- t(m.TR %*% v.e) %*% v.dw # total QALYs for ith simulation
c(TC, TE) # output of the function TC and TE
} # close the Markov function
Results <- sapply(1:n.sim, Markov) # Run the Markov function for n.sim times and save the results in the variable Results
TC2 <- Results[1, ] # save costs separately
TE2 <- Results[2, ] # save QALYs separately
t2 <- Sys.time() - p2 # calculate time to run the analysis by extracting p from the current system time
##############################################################################################
################ probabilistic analysis method 1.0 ###########################################
##############################################################################################
################ generate values outside of the for loop #####################################
p3 <- Sys.time() # save system time
set.seed(1) # set a seed to be able to reproduce the same results
p.HD <- rbeta(n.sim, 20, 980) # probability from healthy to dead
p.HS <- rbeta(n.sim, 50, 950) # probability from healthy to sick
p.SD <- rbeta(n.sim, 100, 900) # probability from sick to dead
c.H <- rnorm(n.sim, 400, 50) # cost of being in healthy state
c.S <- rnorm(n.sim, 100, 80) # cost of being in sick state
c.D <- 0
u.H <- rnorm(n.sim, 0.8, 0.02) # utility of being in healthy state
u.S <- rnorm(n.sim, 0.5, 0.02) # utility of being in the sick state
u.D <- 0 # utility of being dead
TC3 <- TE3 <- vector("numeric", n.sim) # initiate te vector to store costs and effects
m.TR <- array(0, dim = c(n.t + 1, n.s, n.sim)) # initiate the Markov trace
for (i in 1: n.sim){ # start the loop for every simulation
m.P <- matrix(c(1 - p.HD[i] - p.HS[i], p.HS[i], p.HD[i], # transition matrix
0, 1 - p.SD[i], p.SD[i],
0, 0, 1),
byrow = TRUE, nrow = n.s, ncol = n.s,
dimnames = list(v.n, v.n))
m.TR[1, , i] <- c(1, 0, 0) # initialize Markov trace
for (t in 1:n.t) { # throughout the number of cycles
m.TR[t + 1, , i] <- m.TR[t , , i] %*% m.P # estimate the Markov trace for cycle t + 1
} # close the for loop for time
v.c <- c(c.H[i], c.S[i], c.D) # vector of costs
v.e <- c(u.H[i], u.S[i], u.D) # vector of utilities
TC3[i] <- t(m.TR[, , i] %*% v.c) %*% v.dw # calculate total discounted cost for ith simulation
TE3[i] <- t(m.TR[, , i] %*% v.e) %*% v.dw # calculate total discounted QALYs for ith simulation
} # close function
t3 <- Sys.time() - p3 # calculate time to run the analusis by extracting p from the current system time
################ probabilistic analysis method 1.1 ###########################################
################ generate values outside of the Markov function - use sapply #################
p4 <- Sys.time() # save system time
set.seed(1) # set a seed to be able to reproduce the same results
p.HD <- rbeta(n.sim, 20, 980) # probability from healthy to dead
p.HS <- rbeta(n.sim, 50, 950) # probability from healthy to sick
p.SD <- rbeta(n.sim, 100, 900) # probability from sick to dead
c.H <- rnorm(n.sim, 400, 50) # cost of being in healthy state
c.S <- rnorm(n.sim, 100, 80) # cost of being in sick state
c.D <- 0
u.H <- rnorm(n.sim, 0.8, 0.02) # utility of being in healthy state
u.S <- rnorm(n.sim, 0.5, 0.02) # utility of being in the sick state
u.D <- 0
Markov2 <- function(i) { # Markov model function - for different values of index i calculate the Markov model
m.P <- matrix(c(1 - p.HD[i] - p.HS[i], p.HS[i], p.HD[i], # Transition matrix
0, 1 - p.SD[i], p.SD[i],
0, 0, 1),
byrow = T, nrow = n.s,
dimnames = list(v.n, v.n))
m.TR <- matrix(NA, nrow = n.t + 1, ncol = n.s,
dimnames = list(0:n.t, v.n)) # create Markov trace
m.TR[1, ] <- c(1, 0, 0) # initialize Markov trace
for (t in 1:n.t) { # throughout the number of cycles
m.TR[t + 1, ] <- m.TR[t, ] %*% m.P # estimate the Markov trace for cycle t + 1
} # close the for loop for time
v.c <- c(c.H[i], c.S[i], c.D) # vector of costs
v.e <- c(u.H[i], u.S[i], u.D) # vector of utilities
TC4 <- t(m.TR %*% v.c) %*% v.dw # total discounted cost for ith simulation
TE4 <- t(m.TR %*% v.e) %*% v.dw # total discounted QALYs for ith simulation
c(TC4, TE4) # output of the function TC and TE
} # close the function
Results4 <- sapply(1:n.sim, Markov2) # apply the Markov2 functions for n.Sim times and save the results in the variable Results4
TC4 <- Results4[1, ] # save costs seperately
TE4 <- Results4[2, ] # save QALYs seperately
t4 <- Sys.time() - p4 # calculate time to run the analusis by extracting p from the current system time
################ probabilistic analysis method 1.1 ###########################################
####### probabilistic analysis using difference equation method ##################################
p5 <- Sys.time() # save system time
set.seed(1) # set a seed to be able to reproduce the same results
m.H <- m.S <- m.D <- matrix(0, nrow = n.sim, ncol = n.t + 1,
dimnames = list(1:n.sim, 0:n.t)) #initialize state matrices
m.H[, 1] <- 1 # assign every proportion across the simulations to the Healthy state
for (t in 1:n.t) {
m.H[, t + 1] <- m.H[, t] * (1 - p.HD - p.HS) # calculate the prop of cohort in healthy state at time t + 1
m.S[, t + 1] <- m.S[, t] * (1 - p.SD) + m.H[, t] * p.HS # calculate the prop of cohort in sick state at time t + 1
m.D[, t + 1] <- m.D[, t] + m.H[, t] * p.HD + m.S[, t] * p.SD # calculate the prop of cohort in dead state at time t + 1
} # close the loop for time
TE5 <- (u.H * m.H + u.S * m.S + u.D * m.D) %*% v.dw # total discounted cost for all simulations
TC5 <- (c.H * m.H + c.S * m.S + c.D * m.D) %*% v.dw # total discounted QALYs for all simulations
mean(TC5) # calculate the mean discounted costs
mean(TE5) # calculate the mean discounted QALYs
t5 <- Sys.time() - p5 # calculate time to run the analysis by extracting p from the current system time
TC <- cbind(TC1, TC2, TC3, TC4, TC5) # create a variable with the total costs for the different methods
TE <- cbind(TE1, TE2, TE3, TE4, TE5) # create a variable with the total utilities for the different methods
tt <- c(t1, t2, t3, t4, t5) # count the time it took via each method
colMeans(TE) # calculate the column mean of the TE variable
tt # print the time it took for each model to run
TE1 <- data.frame(TE1)
TC1 <- data.frame(TC1)
ScatterCE(strategies = "treatment", m.e = TE1, m.c = TC1) # cost-effectiveness plane
gen_psa <- function(n.sim = 1000, seed = 1){
set.seed(seed) # set a seed to be able to reproduce the same results
df.psa <- data.frame(
p.HD <- rbeta(n.sim, 20, 980) # probability from healthy to dead
p.HS <- rbeta(n.sim, 50, 950) # probability from healthy to sick
p.SD <- rbeta(n.sim, 100, 900) # probability from sick to dead
c.H <- rnorm(n.sim, 400, 50) # cost of being in healthy state
c.S <- rnorm(n.sim, 100, 80) # cost of being in sick state
c.D <- 0
u.H <- rnorm(n.sim, 0.8, 0.02) # utility of being in healthy state
u.S <- rnorm(n.sim, 0.5, 0.02) # utility of being in the sick state
u.D <- 0 # utility of being dead
)
return(df.psa)
}
gen_psa <- function(n.sim = 1000, seed = 1){
set.seed(seed) # set a seed to be able to reproduce the same results
df.psa <- data.frame(
p.HD <- rbeta(n.sim, 20, 980), # probability from healthy to dead
p.HS <- rbeta(n.sim, 50, 950), # probability from healthy to sick
p.SD <- rbeta(n.sim, 100, 900), # probability from sick to dead
c.H <- rnorm(n.sim, 400, 50) , # cost of being in healthy state
c.S <- rnorm(n.sim, 100, 80) , # cost of being in sick state
c.D <- 0,
u.H <- rnorm(n.sim, 0.8, 0.02), # utility of being in healthy state
u.S <- rnorm(n.sim, 0.5, 0.02), # utility of being in the sick state
u.D <- 0 # utility of being dead
)
return(df.psa)
}
gen_psa(10)
gen_psa <- function(n.sim = 1000, seed = 1){
set.seed(seed) # set a seed to be able to reproduce the same results
df.psa <- data.frame(
p.HD = rbeta(n.sim, 20, 980), # probability from healthy to dead
p.HS = rbeta(n.sim, 50, 950), # probability from healthy to sick
p.SD = rbeta(n.sim, 100, 900), # probability from sick to dead
c.H = rnorm(n.sim, 400, 50) , # cost of being in healthy state
c.S = rnorm(n.sim, 100, 80) , # cost of being in sick state
c.D = 0,
u.H = rnorm(n.sim, 0.8, 0.02), # utility of being in healthy state
u.S = rnorm(n.sim, 0.5, 0.02), # utility of being in the sick state
u.D = 0 # utility of being dead
)
return(df.psa)
}
gen_psa(10)
# wrap the code 3-state Markov model in a function
MM.3state <- function(params) {
with(as.list(params),
{
v.dw <- 1 / (1 + d.r) ^ (0:n.t) # calculate discount weight for each cycle based on discount rate d.r
# create a transition matrix
m.P <- matrix(c(1 - p.HD - p.HS, p.HS, p.HD,
0, 1- p.SD, p.SD,
0, 0, 1),
nrow = n.s, ncol = n.s, byrow = TRUE,
dimnames = list(v.n, v.n)) # name the columns and rows of the transition matrix
m.TR <- matrix(NA, nrow = n.t + 1 , ncol = n.s,
dimnames = list(0:n.t, v.n)) # create Markov trace (n.t + 1 because R doesn't understand Cycle 0)
m.TR[1, ] <- c(1, 0, 0) # initialize Markov trace
####### PROCESS ###########################################
for (t in 1:n.t){ # throughout the number of cycles
m.TR[t + 1, ] <- m.TR[t, ] %*% m.P # estimate the Markov trace for cycle the next cycle (t + 1)
}
####### OUTPUT ###########################################
# mean cost and QALYs per cycle
v.c <- m.TR %*% c(c.H, c.S, c.D) # calculate expected costs by multiplying m.TR with the cost vector for the different health states
v.u <- m.TR %*% c(u.H, u.S, u.D) # calculate expected QALYs by multiplying m.TR with the utilities for the different health states
# discounted QALYs and costs
TC <- t(v.c) %*% v.dw # Discount costs by multiplying the cost vector with discount weights (v.dw)
TE <- t(v.u) %*% v.dw # Discount QALYS by multiplying the QALYs vector with discount weights (v.dw)
results <- c(TC, TE)
return(results)
}
)
}
MM.3state(gen_psa(1))
MM.3state(gen_psa(1))
MM.3state(gen_psa(2))
df.psa <- gen_psa(1000)
n.sim <- 1000
n.sim <- 1000
df.psa <- gen_psa(nsim)
n.sim <- 1000
df.psa <- gen_psa(n.sim = nsim)
df.psa <- gen_psa(n.sim = n.sim)
df.out <- matrix(NaN, nrow = n.sim, ncol = 2)
for(i in 1:n.sim){
df.out[i,] <- MM.3state(df.psa[i, ])
cat('\r', paste(round(i/n.sim * 100), "% done", sep = " ")) # display the progress of the simulation
}
# Try it
gen_psa(10) # Works!
# wrap the code 3-state Markov model in a function
MM.3state <- function(params) {
with(as.list(params),
{
v.dw <- 1 / (1 + d.r) ^ (0:n.t) # calculate discount weight for each cycle based on discount rate d.r
# create a transition matrix
m.P <- matrix(c(1 - p.HD - p.HS, p.HS, p.HD,
0, 1- p.SD, p.SD,
0, 0, 1),
nrow = n.s, ncol = n.s, byrow = TRUE,
dimnames = list(v.n, v.n)) # name the columns and rows of the transition matrix
m.TR <- matrix(NA, nrow = n.t + 1 , ncol = n.s,
dimnames = list(0:n.t, v.n)) # create Markov trace (n.t + 1 because R doesn't understand Cycle 0)
m.TR[1, ] <- c(1, 0, 0) # initialize Markov trace
####### PROCESS ###########################################
for (t in 1:n.t){ # throughout the number of cycles
m.TR[t + 1, ] <- m.TR[t, ] %*% m.P # estimate the Markov trace for cycle the next cycle (t + 1)
}
####### OUTPUT ###########################################
# mean cost and QALYs per cycle
v.c <- m.TR %*% c(c.H, c.S, c.D) # calculate expected costs by multiplying m.TR with the cost vector for the different health states
v.u <- m.TR %*% c(u.H, u.S, u.D) # calculate expected QALYs by multiplying m.TR with the utilities for the different health states
# discounted QALYs and costs
TC <- t(v.c) %*% v.dw # Discount costs by multiplying the cost vector with discount weights (v.dw)
TE <- t(v.u) %*% v.dw # Discount QALYS by multiplying the QALYs vector with discount weights (v.dw)
results <- c(TC, TE)
return(results)
}
)
}
p
p <- Sys.time() # save system time
for(i in 1:n.sim){
df.out[i,] <- MM.3state(df.psa[i, ])
cat('\r', paste(round(i/n.sim * 100), "% done", sep = " ")) # display the progress of the simulation
}
t.psa <- Sys.time() - p # calculate time to run the analusis by extracting p from the current system time
t.psa
ScatterCE(strategies = "treatment", m.e = TE, m.c = TC) # cost-effectiveness plane
ScatterCE
TE
ScatterCE(strategies = "treatment", m.e = df.out[, "TE"], m.c = df.out[, "TC"]) # cost-effectiveness plane
## Initialize matrix of outcomes
df.out <- matrix(NaN, nrow = n.sim, ncol = 2)
colnames(df.out) <- c("TE", "TC")
## Run PSA
p <- Sys.time() # save system time
for(i in 1:n.sim){
df.out[i,] <- MM.3state(df.psa[i, ])
cat('\r', paste(round(i/n.sim * 100), "% done", sep = " ")) # display the progress of the simulation
}
t.psa <- Sys.time() - p # calculate time to run the analusis by extracting p from the current system time
t.psa
p <- Sys.time() # save system time
for(i in 1:n.sim){
df.out[i,] <- MM.3state(df.psa[i, ])
cat('\r', paste(round(i/n.sim * 100), "% done", sep = " ")) # display the progress of the simulation
}
t.psa <- Sys.time() - p # calculate time to run the analusis by extracting p from the current system time
t.psa
ScatterCE(strategies = "treatment", m.e = df.out[, "TE"], m.c = df.out[, "TC"]) # cost-effectiveness plane
df.out[, "TE"]
df.out[, "TC"]
## Initialize matrix of outcomes
df.out <- as.data.framd(matrix)(NaN, nrow = n.sim, ncol = 2)
colnames(df.out) <- c("TE", "TC")
## Run PSA
p <- Sys.time() # save system time
for(i in 1:n.sim){
df.out[i,] <- MM.3state(df.psa[i, ])
cat('\r', paste(round(i/n.sim * 100), "% done", sep = " ")) # display the progress of the simulation
}
t.psa <- Sys.time() - p # calculate time to run the analusis by extracting p from the current system time
t.psa
ScatterCE(strategies = "treatment", m.e = df.out[, "TE"], m.c = df.out[, "TC"]) # cost-effectiveness plane
str(df.out)
str(df.out[, "TE"])
ScatterCE(strategies = "treatment", m.e = matrix(df.out[, "TE"]), m.c = matrix(df.out[, "TC"])) # cost-effectiveness plane
ScatterCE(strategies = "treatment", m.e = data.frame(df.out[, "TE"]), m.c = data.frame(df.out[, "TC"])) # cost-effectiveness plane
rm(list = ls()) # Delete everything that is in R's memory
setwd("/Users/fernando/Google Drive/Projects/DARTH/Courses/2018 Merck/Participants's material/Day 2/") # set the working directory
source("Functions/PSA_functions.R") # load custom made functions for PSA
######################### INPUT PARAMETERS ####################################
v.n <- c("Healthy", "Sick", "Dead") # state names
n.s <- length(v.n) # number of states
n.t <- 60 # number of cycles
n.sim <- 10000 # number of simulations
d.r <- 0.03 # discount rate
v.dw <- 1 / (1 + d.r) ^ (0:n.t) # calculate discount weight for each cycle based on discount rate d.r
## Function that generates random sample for PSA
gen_psa <- function(n.sim = 1000, seed = 1){
set.seed(seed) # set a seed to be able to reproduce the same results
df.psa <- data.frame(
p.HD = rbeta(n.sim, 20, 980), # probability from healthy to dead
p.HS = rbeta(n.sim, 50, 950), # probability from healthy to sick
p.SD = rbeta(n.sim, 100, 900), # probability from sick to dead
c.H = rnorm(n.sim, 400, 50) , # cost of being in healthy state
c.S = rnorm(n.sim, 100, 80) , # cost of being in sick state
c.D = 0,
u.H = rnorm(n.sim, 0.8, 0.02), # utility of being in healthy state
u.S = rnorm(n.sim, 0.5, 0.02), # utility of being in the sick state
u.D = 0 # utility of being dead
)
return(df.psa)
}
# Try it
gen_psa(10) # Works!
# wrap the code 3-state Markov model in a function
MM.3state <- function(params) {
with(as.list(params),
{
v.dw <- 1 / (1 + d.r) ^ (0:n.t) # calculate discount weight for each cycle based on discount rate d.r
# create a transition matrix
m.P <- matrix(c(1 - p.HD - p.HS, p.HS, p.HD,
0, 1- p.SD, p.SD,
0, 0, 1),
nrow = n.s, ncol = n.s, byrow = TRUE,
dimnames = list(v.n, v.n)) # name the columns and rows of the transition matrix
m.TR <- matrix(NA, nrow = n.t + 1 , ncol = n.s,
dimnames = list(0:n.t, v.n)) # create Markov trace (n.t + 1 because R doesn't understand Cycle 0)
m.TR[1, ] <- c(1, 0, 0) # initialize Markov trace
####### PROCESS ###########################################
for (t in 1:n.t){ # throughout the number of cycles
m.TR[t + 1, ] <- m.TR[t, ] %*% m.P # estimate the Markov trace for cycle the next cycle (t + 1)
}
####### OUTPUT ###########################################
# mean cost and QALYs per cycle
v.c <- m.TR %*% c(c.H, c.S, c.D) # calculate expected costs by multiplying m.TR with the cost vector for the different health states
v.u <- m.TR %*% c(u.H, u.S, u.D) # calculate expected QALYs by multiplying m.TR with the utilities for the different health states
# discounted QALYs and costs
TC <- t(v.c) %*% v.dw # Discount costs by multiplying the cost vector with discount weights (v.dw)
TE <- t(v.u) %*% v.dw # Discount QALYS by multiplying the QALYs vector with discount weights (v.dw)
results <- c(TC, TE)
return(results)
}
)
}
## Draw random sample for PSA
df.psa <- gen_psa(n.sim = n.sim)
## Initialize matrix of outcomes
df.out <- matrix(NaN, nrow = n.sim, ncol = 2)
colnames(df.out) <- c("TE", "TC")
## Run PSA
p <- Sys.time() # save system time
for(i in 1:n.sim){
df.out[i,] <- MM.3state(df.psa[i, ])
cat('\r', paste(round(i/n.sim * 100), "% done", sep = " ")) # display the progress of the simulation
}
t.psa <- Sys.time() - p # calculate time to run the analusis by extracting p from the current system time
t.psa
ScatterCE(strategies = "treatment", m.e = data.frame(df.out[, "TE"]), m.c = data.frame(df.out[, "TC"])) # cost-effectiveness plane
rm(list = ls()) # Delete everything that is in R's memory
setwd("/Users/fernando/Google Drive/Projects/DARTH/Courses/2018 Merck/Participants's material/Day 2/") # set the working directory
source("Functions/PSA_functions.R") # load custom made functions for PSA
######################### INPUT PARAMETERS ####################################
v.n <- c("Healthy", "Sick", "Dead") # state names
n.s <- length(v.n) # number of states
n.t <- 60 # number of cycles
n.sim <- 10000 # number of simulations
d.r <- 0.03 # discount rate
v.dw <- 1 / (1 + d.r) ^ (0:n.t) # calculate discount weight for each cycle based on discount rate d.r
####### probabilistic analysis ###########################################
####### Create function that generate random samples ###################################
## Function that generates random sample for PSA
gen_psa <- function(n.sim = 1000, seed = 1){
set.seed(seed) # set a seed to be able to reproduce the same results
df.psa <- data.frame(
p.HD = rbeta(n.sim, 20, 980), # probability from healthy to dead
p.HS = rbeta(n.sim, 50, 950), # probability from healthy to sick
p.SD = rbeta(n.sim, 100, 900), # probability from sick to dead
c.H = rnorm(n.sim, 400, 50) , # cost of being in healthy state
c.S = rnorm(n.sim, 100, 80) , # cost of being in sick state
c.D = 0,
u.H = rnorm(n.sim, 0.8, 0.02), # utility of being in healthy state
u.S = rnorm(n.sim, 0.5, 0.02), # utility of being in the sick state
u.D = 0 # utility of being dead
)
return(df.psa)
}
# Try it
gen_psa(10) # Works!
# wrap the code 3-state Markov model in a function
MM.3state <- function(params) {
with(as.list(params),
{
v.dw <- 1 / (1 + d.r) ^ (0:n.t) # calculate discount weight for each cycle based on discount rate d.r
# create a transition matrix
m.P <- matrix(c(1 - p.HD - p.HS, p.HS, p.HD,
0, 1- p.SD, p.SD,
0, 0, 1),
nrow = n.s, ncol = n.s, byrow = TRUE,
dimnames = list(v.n, v.n)) # name the columns and rows of the transition matrix
m.TR <- matrix(NA, nrow = n.t + 1 , ncol = n.s,
dimnames = list(0:n.t, v.n)) # create Markov trace (n.t + 1 because R doesn't understand Cycle 0)
m.TR[1, ] <- c(1, 0, 0) # initialize Markov trace
####### PROCESS ###########################################
for (t in 1:n.t){ # throughout the number of cycles
m.TR[t + 1, ] <- m.TR[t, ] %*% m.P # estimate the Markov trace for cycle the next cycle (t + 1)
}
####### OUTPUT ###########################################
# mean cost and QALYs per cycle
v.c <- m.TR %*% c(c.H, c.S, c.D) # calculate expected costs by multiplying m.TR with the cost vector for the different health states
v.u <- m.TR %*% c(u.H, u.S, u.D) # calculate expected QALYs by multiplying m.TR with the utilities for the different health states
# discounted QALYs and costs
TC <- t(v.c) %*% v.dw # Discount costs by multiplying the cost vector with discount weights (v.dw)
TE <- t(v.u) %*% v.dw # Discount QALYS by multiplying the QALYs vector with discount weights (v.dw)
results <- c(TC, TE)
return(results)
}
)
}
## Draw random sample for PSA
df.psa <- gen_psa(n.sim = n.sim)
## Initialize matrix of outcomes
df.out <- matrix(NaN, nrow = n.sim, ncol = 2)
colnames(df.out) <- c("TE", "TC")
## Run PSA
p <- Sys.time() # save system time
for(i in 1:n.sim){
df.out[i,] <- MM.3state(df.psa[i, ])
cat('\r', paste(round(i/n.sim * 100), "% done", sep = " ")) # display the progress of the simulation
}
t.psa <- Sys.time() - p # calculate time to run the analusis by extracting p from the current system time
t.psa
ScatterCE(strategies = "treatment", m.e = data.frame(df.out[, "TE"]), m.c = data.frame(df.out[, "TC"])) # cost-effectiveness plane
#install.packages(c("reshape2"), dependencies = TRUE)
setwd(dirname(rstudioapi::getActiveDocumentContext()$path))
source("../Functions/PSA_functions.R") # load custom made functions for PSA
## Clean everything from workspace
rm(list=ls())
setwd(dirname(rstudioapi::getActiveDocumentContext()$path))
source("../Functions/PSA_functions.R")
source("../Functions/DARTH_functions.R")
## Load PSA file
# Read the `.csv` simulation file into `R`.
psa <- read.csv("PSA/PSA.csv",header=TRUE)
## Create initial parameters
# Label our strategies
Strategies <- c("Chemo", "Radio", "Surgery")
## Load PSA file
# Read the `.csv` simulation file into `R`.
psa <- read.csv("PSA.csv",header=TRUE)
## Create initial parameters
# Label our strategies
Strategies <- c("Chemo", "Radio", "Surgery")
Strategies
#Determine the number of strategies
numDepVars <- length(Strategies)
numDepVars
## Create matrices for outcomes and parameters
Outcomes <- psa[, 2:(numDepVars*2+1)]
head(Outcomes)
#######################################
#### Displaying results from a PSA ####
#######################################
### Define matrices with costs and effects
m.c <- Outcomes[, c(1, 3, 5)]
m.e <- Outcomes[, c(2, 4, 6)]
### Define a vector of WTP values
v.wtp <- c(1, seq(5000, 150000, length.out = 31))
### CE Plane and Scatter Plot ####
library(plyr) # To summarize data and generate new data frames from it
library(ggplot2)
ce.mat <- cbind(Strategies = 1:3,
Costs = colMeans(m.c),
Effectviness = colMeans(m.e))
ce.front <- getFrontier(ce.mat, plot = F)
ce.df <- data.frame(Strategy = Strategies,
Cost = colMeans(m.c),
Effectiveness = colMeans(m.e))
plotFrontier(CEmat = ce.df, frontier = ce.front)
### With user-written function & ggplot2
ScatterCE(strategies = Strategies,
m.c = m.c,
m.e = m.e)
#### CEA Curves (CEAC) & Frontier (CEAF) ####
ceaf(v.wtp = v.wtp,
strategies = Strategies,
m.e = m.e,
m.c = m.c)
#### Expected Loss Curves (ELC) ####
elc(v.wtp = v.wtp,
strategies = Strategies,
m.e = m.e,
m.c = m.c)
#### Expected Loss Curves (ELC) ####
evpi(v.wtp = v.wtp,
m.e = m.e,
m.c = m.c)