forked from Open-Systems-Pharmacology/Inhalation-model
-
Notifications
You must be signed in to change notification settings - Fork 0
/
populate_model.R
586 lines (456 loc) · 23.5 KB
/
populate_model.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
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
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
####################
#
# File name : populate_model.R
# Author : Moriah Pellowe
# Date created : September 10, 2020
#
# This script will:
# - calculate the deposition fractions in R
# - populate the corresponding parameters into a MoBi pkml file
# - save the simulation
#
# NOTE:
# - the number of bins (or particle diameters) and the molecule name should match what is in the MoBi pkml file
#
# TIPS:
# - getSimulationTree() to get proper file path
# - paste() to break down long lines of code
#
####################
populate_model <- function(pkml_file, molecule_name, particle_diameters_dm, mean_particle_radius_dm, sd_particle_radius_dm,
oral_bioavailability = 1, lung_bioavailability = 1, device_bioavailabliity = 1, logScale = FALSE,
breathing_frequency_N_min = 15, fraction_inspiratory = 0.5, breath_hold_time_sec = 0,
delay_volume_mL = 0, tidal_volume_mL = 1000, bolus_volume_mL = 1000) {
# Load in libraries and scripts
library(ospsuite)
#source("deposition_interface_v2.R")
# load in lung model as pkml
sim <- loadSimulation(pkml_file)
### PARAMETERS: Define arguments for deposition function ###
# particle_diameters_dm <- c(2.1e-5) #c(1.35e-5, 2.1e-5, 2.85e-5) #c(4e-5, 1e-4, 1.6e-4)
# mean_particle_radius_dm <- 2.1e-5 #1e-4
# sd_particle_radius_dm <- 0.75e-5 #3e-5
# logScale <- FALSE
# molecule_name <- "Salbutamol"
# initialization
numberOfBins <- length(particle_diameters_dm)
# read in drug density from simulation
density_kg_dm3 <- getParameter(paste(molecule_name, "|Density (drug)", sep=""), sim)
drug_density_kg_m3 <- density_kg_dm3$value*1000
# calculate deposition fractions
deposition_output <- deposition_interface(particle_diameters_dm, mean_particle_radius_dm, sd_particle_radius_dm, drug_density_kg_m3, logScale,
breathing_frequency_N_min, fraction_inspiratory, breath_hold_time_sec, delay_volume_mL, tidal_volume_mL, bolus_volume_mL)
# adjust the deposition fractions
# note that the oral bioavailability is changed within the MoBi simulation so it is not accounted for here
deposition_output$distribution_across_gens[1,] <- device_bioavailabliity*deposition_output$distribution_across_gens[1,]
# absolute bioavailability after inhaled administration with oral charcoal
# = fraction output by device * fraction of drug deposited in lung * lung bioavailability
# i.e. F_inh,charcoal = F_device * df_lung * F_lung
deposition_output$distribution_across_gens[2:dim(deposition_output$distribution_across_gens)[1],] <-
device_bioavailabliity*deposition_output$distribution_across_gens[2:dim(deposition_output$distribution_across_gens)[1],]*lung_bioavailability
# set oral bioavailability
paths <- "Organism|ExtrathoracicRegion|Oral bioavailability - F_oral"
setParameterValuesByPath(paths, oral_bioavailability, sim)
# set particle radii
paths <- NULL
for (bin in 1:numberOfBins) {
temp <- paste("Applications|Administration Protocol - Lung|Formulation - Particle Dissolution - Polydisperse|Application_1|ParticleBin_",
toString(bin), "|Particle radius (at t=0)", sep="")
paths <- c(paths, temp)
}
particle_radius_dm <- particle_diameters_dm/2
setParameterValuesByPath(paths, particle_radius_dm, sim)
# set Number_Of_Particles_Factor
paths <- NULL
for (bin in 1:numberOfBins) {
temp <- paste("Applications|Administration Protocol - Lung|Formulation - Particle Dissolution - Polydisperse|Application_1|ParticleBin_",
toString(bin), "|Number_Of_Particles_Factor", sep="")
paths <- c(paths, temp)
}
setParameterValuesByPath(paths, deposition_output$number_of_particles_factor, sim)
# set generations with slices
paths <- NULL
values <- NULL
gens_w_slices <- data.frame("Generation" = c(1:2,5:16), "Slices" = c(2,2,3,4,6,7,9,12,14,17,19,21,22,24))
for (bin in 1:numberOfBins) {
for (index in 1:nrow(gens_w_slices)) {
gen <- gens_w_slices$Generation[index]
numberOfSlices <- gens_w_slices$Slices[index]
for (slice in 1:numberOfSlices) {
# add zero to generation or slice if single digit value
genZero <- ifelse(gen < 10, toString(0), "")
sliceZero <- ifelse(slice < 10, toString(0), "")
temp <- paste("Applications|Administration Protocol - Lung|Formulation - Particle Dissolution - Polydisperse|Application_1|ParticleBin_",
toString(bin), "|Generation ", genZero, toString(gen), " - Slice ", sliceZero, toString(slice), sep="")
paths <- c(paths, temp)
# note that the generation is off by one because 1 corresponds to ET and i+1 corresponds to generation i for row > 1
values <- c(values, deposition_output$distribution_across_gens[gen+1,bin]/numberOfSlices)
}
}
}
setParameterValuesByPath(paths, values, sim)
# set ET region
paths <- NULL
values <- NULL
for (bin in 1:numberOfBins){
temp <- paste("Applications|Administration Protocol - Lung|Formulation - Particle Dissolution - Polydisperse|Application_1|ParticleBin_",
toString(bin), "|Number of particles fraction - Extrathoracic", sep="")
paths <- c(paths, temp)
}
setParameterValuesByPath(paths, deposition_output$distribution_across_gens[1,], sim)
# set generations without slices
paths <- NULL
values <- NULL
gens_wo_slices <- c(3:4, 17:24)
for (bin in 1:numberOfBins) {
for (gen in gens_wo_slices) {
# add zero to generation or slice if single digit value
genZero <- ifelse(gen < 10, toString(0), "")
temp <- paste("Applications|Administration Protocol - Lung|Formulation - Particle Dissolution - Polydisperse|Application_1|ParticleBin_",
toString(bin), "|Number of particles fraction - Generation ", genZero, toString(gen), sep="")
paths <- c(paths, temp)
# note that the generation is off by one because 1 corresponds to ET and i+1 corresponds to generation i for row > 1
values <- c(values, deposition_output$distribution_across_gens[gen+1,bin])
}
}
setParameterValuesByPath(paths, values, sim)
par <- getParameter(paths[1], sim)
par$value
saveSimulation(sim, paste("populated_", pkml_file, sep=""))
return(deposition_output)
}
#####
#
# Filename: deposition_interface.R
# Author: Moriah Pellowe
# Date created: March 9, 2020
#
# v2: September 30, 2020
# - if statement added so that beta is in proper format for case of 1 particle bin
#
# This script will calculate the number of particles per L of drug volume, the pdf_particles over the diameters and 24 generations,
# as well as the distribution over the generations for each particle diameter.
#
#####
deposition_interface <- function(particle_diameters_dm, mean_particle_radius_dm, sd_particle_radius_dm, drug_density_kg_m3, log_flag=FALSE,
breathing_frequency_N_min, fraction_inspiratory, breath_hold_time_sec,
delay_volume_mL, tidal_volume_mL, bolus_volume_mL){
library(pracma)
#particle_diameters_dm <- c(10^(-9:-5))
#mean_particle_diameter <- 1.5e-6
#sd_particle_diameter <- 6e-7
## Parameters
breath_f_br <- breathing_frequency_N_min # breathing frequency, [1/min]
breath_fr_in <- fraction_inspiratory # fraction of breath as inspiratory
breath_t_b <- breath_hold_time_sec # breath-hold time [s]
numGens <- 24
numSizes <- length(particle_diameters_dm)
breath_V_D <- delay_volume_mL # Delay volume [mL]
breath_V_T <- tidal_volume_mL # Tidal volume [mL]
breath_V_B <- bolus_volume_mL # Bolus volume [mL]
# functional residual capacity is hard-coded since the Weibel structure is scaled to this value of 3000 mL
breath_FRC <- 3000 # Functional residual capacity, FRC [mL]
data_V_lung_tot = 3000
data_alv_frac_value = c(0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0.002, 0.007, 0.02, 0.07, 0.139, 0.282, 0.48)
## Initialization
N <- c(2^(0:(numGens-1)))
stk_in <- matrix(0, nrow=numGens, ncol=numSizes)
stk_exp <- matrix(0, nrow=numGens, ncol=numSizes)
IMP_in <- matrix(0, nrow=numGens, ncol=numSizes)
IMP_exp <- matrix(0, nrow=numGens, ncol=numSizes)
IMP_b <- matrix(0, nrow=numGens, ncol=numSizes)
eps_in <- matrix(0, nrow=numGens, ncol=numSizes)
eps_exp <- matrix(0, nrow=numGens, ncol=numSizes)
SED_in <- matrix(0, nrow=numGens, ncol=numSizes)
SED_exp <- matrix(0, nrow=numGens, ncol=numSizes)
SED_b <- matrix(0, nrow=numGens, ncol=numSizes)
re_in <- matrix(0, nrow=numGens, ncol=numSizes)
re_exp <- matrix(0, nrow=numGens, ncol=numSizes)
delta_in <- matrix(0, nrow=numGens, ncol=numSizes)
delta_exp <- matrix(0, nrow=numGens, ncol=numSizes)
DIF_in <- matrix(0, nrow=numGens, ncol=numSizes)
DIF_exp <- matrix(0, nrow=numGens, ncol=numSizes)
DIF_b <- matrix(0, nrow=numGens, ncol=numSizes)
particle_diameters_m <- particle_diameters_dm/10
particle_radius_dm <- particle_diameters_dm/2
## Lung morphometry [dm]
D <- c(0.1539, 0.1043, 0.071, 0.0479, 0.0385, 0.0299, 0.0239, 0.0197, 0.0159,
0.0132, 0.0111, 0.0093, 0.0081, 0.007, 0.0063, 0.0056, 0.0051, 0.0046,
0.0043, 0.004, 0.0038, 0.0037, 0.0035, 0.0035)
L <- c(1.026, 0.407, 0.1624, 0.065, 0.1086, 0.0915, 0.0769, 0.065, 0.0547,
0.0462, 0.0393, 0.0333, 0.0282, 0.0231, 0.0197, 0.0171, 0.0141, 0.0121,
0.01, 0.0085, 0.0071, 0.006, 0.005, 0.0043)
V <- (L*10)*N*pi*(D*10/2)^2 # [cm3]
# Scale alveolar region
V_add <- data_V_lung_tot - sum(V)
V <- V + c(data_alv_frac_value)*V_add
phi <- rep((pi/4),length(D))
## BolusScaling.m
TLC <- sum(V)
scale_init <- breath_FRC/TLC
Vi_init <- V*scale_init
cum_V <- cumsum(Vi_init)
cum_Vi_init <- sum(Vi_init)*rep(1, length(Vi_init))- c(0, cum_V[1:(length(cum_V)-1)])
scale_f <- rep(1, length(Vi_init))
for (i in 1:length(Vi_init)) {
scale_f[i] <- (sum(Vi_init) + breath_V_D)/cum_Vi_init[i]
}
scale_t <- rep(1, length(Vi_init))
alpha <- rep(0, length(Vi_init))
for (i in 1:length(Vi_init)) {
# Special treatment for the last generation
if (i == length(Vi_init)) {
alpha[i] <- sum(Vi_init[i:length(Vi_init)])
scale_t[i] <- (sum(Vi_init) + breath_V_D + breath_V_B)/alpha[i]
} else {
alpha[i] <- sum(Vi_init[(i+1):length(Vi_init)])
scale_t[i] <- (sum(Vi_init) + breath_V_D + breath_V_B)/alpha[i]
}
}
checking <- rep(0, length(Vi_init)-1)
for (i in (1:(length(Vi_init)-1))){
checking[i] <- (breath_V_T < ((cum_V[i] + breath_V_D + breath_V_B)/(1 - cum_V[i]/sum(Vi_init)))) # This formula accounts for scaling
}
# Check if bolus is not washed out in mouth (before trachea)
if (breath_V_T < (breath_V_D + breath_V_B)){
i_wash <- length(Vi_init)
} else {
i_wash <- which.max(checking)
}
scale_t[i_wash:length(scale_t)] <- (sum(Vi_init) + breath_V_T)/sum(Vi_init)
f_ave <- (scale_f + scale_t)/2
L <- L*f_ave^(1/3)
D <- D*f_ave^(1/3)
D_m <- D/10
L_m <- L/10
V_scaled <- f_ave*Vi_init
V_cum_scaled <- cumsum(V_scaled)
checking2 <- breath_V_T - breath_V_D < (V_cum_scaled[1:(length(V_cum_scaled)-1)])
# If checking2 is false for every generation
if (sum(checking2)==0) {
imax <- length(Vi_init)
} else {
imax <- which.max(checking2)
}
V_scaled[imax] <- (breath_V_T - breath_V_D) - V_cum_scaled[imax-1]
if (imax < length(Vi_init)){
V_scaled[(imax+1):length(V_scaled)] <- Vi_init[(imax+1):length(Vi_init)]
}
V <- V_scaled
V_cum_scaled_updated <- cumsum(V_scaled)
## Constants - extrathoracic deposition
lamda <- 0.066e-6 # 0.066 um, mean free path of air molecule [m]
k <- 1.38064852E-23 # Boltzmann constant [m^2·kg/(s^2·K)]
T <- 310.65 # 37.5 degree Celsius in Kelvin
ne <- 1.9224364E-05 # viscosity of air at 37.5 dgr C [kg/m/s] (1.846*10^-5 kg/m/s at 300K)
## Constants - inertial impaction
po <- drug_density_kg_m3 # Unit particle density, 1 g/cm3 = 1000 kg/m3
## Constants - gravitational sedimentation
g <- 9.81 # gravitational acceleration [m/s^2]
## Constants - diffusion
pa = 1.1372; # density of air 37.5 degr C [kg/m3]; http://www.gribble.org/cycling/air_density.html
## Formulas - extrathoracic deposition
breath_t_in <- breath_fr_in*(1/breath_f_br) # min
breath_t_exp <- (1-breath_fr_in)*(1/breath_f_br) # min
# Formulas - inertial impaction
Cd <- 1 + (lamda/particle_diameters_m)*(2.514 + 0.8*exp(-0.55*(particle_diameters_m/lamda)))
# Flow rate (inspiratory and expiratory)
Q_in <- (breath_V_T/breath_t_in)/1000 # L/min
Q_exp <- (breath_V_T/breath_t_exp)/1000 # L/min
# Brownian diffusion coefficient
Dmol <- ((k*T*Cd)/(3*pi*ne*particle_diameters_m)) # (m^2)/s
Dmol_cm2_s <- Dmol*(100^2)
## Formulas - inertial impaction
Q_in_gen_i <- (Q_in*1000/60)/N # [cm3/s]
Q_exp_gen_i <- (Q_exp*1000/60)/N # [cm3/s]
A_gen_i <- pi*(D*10/2)^2 # [cm2] A=r^2*pi
v_in_gen_i <- (Q_in_gen_i/A_gen_i)/100 # [m/s]
v_exp_gen_i <- (Q_exp_gen_i/A_gen_i)/100 # [m/s]
theta <- L/(4*D)
## Formulas - Gravitational sedimentation
vg <- (po*((particle_diameters_m^2)*g*Cd)/(18*ne)) # Gravitational settling velocity of a particle
t_i_in <- V/N/Q_in_gen_i # [s]
t_i_exp <- V/N/Q_exp_gen_i # [s]
## Extrathoracic deposition
oral_in <- 1 - exp(-0.000278*Q_in *(particle_diameters_m*1e6)^2 - 20.4*(Dmol_cm2_s)^0.66*Q_in ^(-0.31))
oral_exp <- 1 - exp(-0.000278*Q_exp*(particle_diameters_m*1e6)^2 - 20.4*(Dmol_cm2_s)^0.66*Q_exp^(-0.31))
# print(oral_in)
# print(oral_exp)
for(j in 1:numSizes){
for (i in 1:numGens) {
r_i <- D_m[i]/2 # radius of tube
r_i_p <- particle_diameters_m[j]/2 # radius of particle
## Inertial impaction
stk_in[i,j] <- po*(particle_diameters_m[j]^2)*v_in_gen_i[i]*Cd[j]/(9*ne*D_m[i]) # With cunningham, Zhang et al. 1997
stk_exp[i,j] <- po*(particle_diameters_m[j]^2)*v_exp_gen_i[i]*Cd[j]/(9*ne*D_m[i]) # With cunningham, Zhang et al. 1997
IMP_in[i,j] <- 0.768*theta[i]*stk_in[i,j]
IMP_exp[i,j] <- 0.768*theta[i]*stk_exp[i,j]
## Gravitational sedimentation
eps_in[i,j] <- 3*vg[j]*t_i_in[i]*cos(phi[i])/(4*D_m[i]) # upright position
eps_exp[i,j] <- 3*vg[j]*t_i_exp[i]*cos(phi[i])/(4*D_m[i]) # upright position
# to handle case when eps > 1, since it is an argument of asin(x)
if (eps_in[i,j] > 1) {
SED_in[i,j] <- 1 # SEE FEB 10,11 NOTES
} else {
SED_in[i,j] <-
2/pi*(2*eps_in[i,j]*(1-eps_in[i,j]^(2/3))^(1/2) - (eps_in[i,j]^(1/3))*(1-eps_in[i,j]^(2/3))^(1/2) + asin(eps_in[i,j]^(1/3)))
}
if (eps_exp[i,j] > 1) {
SED_exp[i,j] <- 1 # SEE FEB 10, 11 NOTES
} else {
SED_exp[i,j] <-
2/pi*(2*eps_exp[i,j]*(1-eps_exp[i,j]^(2/3))^(1/2) - (eps_exp[i,j]^(1/3))*(1-eps_exp[i,j]^(2/3))^(1/2) + asin(eps_exp[i,j]^(1/3)))
}
# Breath hold
SED_b[i,j] <- 1 - exp( (-4*g*Cd[j]*(r_i_p^2)*breath_t_b*cos(phi[i]))/(9*pi*ne*r_i) )
## Diffusion
delta_in[i,j] <- Dmol[j]*L_m[i]/(v_in_gen_i[i]*(D_m[i]^2))
delta_exp[i,j] <- Dmol[j]*L_m[i]/(v_exp_gen_i[i]*(D_m[i]^2))
re_in[i,j] <- pa*D_m[i]*v_in_gen_i[i]/ne
re_exp[i,j] <- pa*D_m[i]*v_exp_gen_i[i]/ne
if (re_in[i,j] > 2000) { # turbulent flow
DIF_in[i,j] <- 4*(delta_in[i,j]^0.5) * (1 - 0.444*(delta_in[i,j]^0.5)) # the eq. continues with a (...) in Yu and Diu 1982
} else { # laminar flow
DIF_in[i,j] <- 1-0.819*exp(-14.63*delta_in[i,j])-0.0976*exp(-89.22*delta_in[i,j])-0.0325*exp(-228*delta_in[i,j]) - 0.0509*exp(-125.9*delta_in[i,j]^(2/3))
}
if (re_exp[i,j] > 2000) { # turbulent flow
DIF_exp[i,j] <- 4*(delta_exp[i,j]^0.5) * (1 - 0.444*(delta_exp[i,j]^0.5)) # the eq. continues with a (...) in Yu and Diu 1982
} else { # laminar flow
DIF_exp[i,j] <- 1-0.819*exp(-14.63*delta_exp[i,j])-0.0976*exp(-89.22*delta_exp[i,j])-0.0325*exp(-228*delta_exp[i,j]) - 0.0509*exp(-125.9*delta_exp[i,j]^(2/3))
}
# Breath-hold
DIF_b[i,j] <- 1 - exp(-5.784*k*T*Cd[j]*breath_t_b/(6*pi*ne*r_i_p*(r_i^2)))
}
}
# print(IMP_in)
# print(IMP_exp)
# Remove any complex numbers from asin(x), only necessary when eps > 1
#SED_in <- Re(SED_in)
#SED_exp <- Re(SED_exp)
# print(SED_in)
# print(SED_exp)
# print(SED_b)
# print(DIF_in)
# print(DIF_exp)
# print(DIF_b)
# Add row at top of each matrix to represent mouth (before trachea)
IMP_in <- rbind(oral_in, IMP_in)
IMP_exp <- rbind(oral_exp, IMP_exp)
IMP_b <- rbind(rep(0,length(oral_in)), IMP_b)
SED_in <- rbind(rep(0,length(oral_in)), SED_in)
SED_exp <- rbind(rep(0,length(oral_in)), SED_exp)
SED_b <- rbind(rep(0,length(oral_in)), SED_b)
DIF_in <- rbind(rep(0,length(oral_in)), DIF_in)
DIF_exp <- rbind(rep(0,length(oral_in)), DIF_exp)
DIF_b <- rbind(rep(0,length(oral_in)), DIF_b)
V_cum_scaled_updated <- c(0, V_cum_scaled_updated)
V_scaled <- c(0, V_scaled)
# Compensate for extra row
i_wash <- i_wash+1
imax <- imax+1
## ApplyFractions.m
numGens_dep <- imax
f <- matrix(0, nrow=(numGens_dep+1), ncol=numSizes)
DEP_in <- matrix(0, nrow=numGens_dep, ncol=numSizes)
DEP_exp <- matrix(0, nrow=numGens_dep, ncol=numSizes)
DEP_b <- matrix(0, nrow=numGens_dep, ncol=numSizes)
P_in <- 1-(1-IMP_in)*(1-SED_in)*(1-DIF_in)
P_exp <- 1-(1-IMP_exp)*(1-SED_exp)*(1-DIF_exp)
P_b <- 1-(1-IMP_b)*(1-SED_b)*(1-DIF_b)
for (j in 1:numSizes){
f[,j] <- cumprod(c(1, (1-P_in[1:numGens_dep,j])))
}
# BolusScaling.m
frac_bolus <- rep(0, numGens+1)
frac_bolus[1:(i_wash-1)] <- rep(1, (i_wash-1))
VF_cum <- rep(0, numGens+1)
for (i in i_wash:imax) {
VF_cum[i] = (breath_V_T - breath_V_D - V_cum_scaled_updated[i-1])/breath_V_B
}
VF_cum[VF_cum>1] <- 1
frac_bolus[i_wash:imax] <- VF_cum[i_wash:imax]
for (j in 1:numSizes){
DEP_in[,j] <- f[1:(dim(f)[1]-1),j]*P_in[1:imax,j]*frac_bolus[1:numGens_dep]
DEP_b[,j] <- f[1:(dim(f)[1]-1),j]*(1-P_in[1:imax,j])*P_b[1:imax,j]*V_scaled[1:numGens_dep]
}
# print(DEP_in)
# print(DEP_b)
# frac_pause -> BolusScaling.m
frac_pause <- rep(0, length(frac_bolus))
for (i in 2:imax) {
if (i < imax) {
frac_pause[i] <- frac_bolus[i] - frac_bolus[i+1]
} else if (i == imax) {
frac_pause[i] <- 1 - sum(frac_pause)
}
}
frac_pause_matrix <- matrix(rep(frac_pause[1:imax], times=numSizes), ncol=numSizes)
x <- matrix(0, nrow=numGens_dep, ncol=numSizes)
if (numSizes==1) {
beta <- array(f[3:dim(f)[1],]*frac_pause_matrix[2:dim(frac_pause_matrix)[1],], dim = c(dim(f)[1]-2,1))
} else {
beta <- f[3:dim(f)[1],]*frac_pause_matrix[2:dim(frac_pause_matrix)[1],]
}
for (i in (numGens_dep-1):1) {
x[i,] <- (1-P_exp[i+1,])*x[i+1,] + (1-P_b[i+1,])*beta[i,] #*(1-P_b(i+1,:))
}
DEP_exp <- x*P_exp[1:imax,]
# print(DEP_exp)
# add up the deposition fractions
total_deposition <- DEP_in + DEP_exp + DEP_b
# add rows for generations with no deposition
if (dim(total_deposition)[1] < (numGens+1)) {
total_deposition <- rbind(total_deposition, matrix(0, nrow=(numGens+1-dim(total_deposition)[1]),
ncol = dim(total_deposition)[2]))
}
# Calculate proportion of particles based on radius (in dm) # NOTE: Boger did this in decimetres
if (log_flag) {
# convert geometric mean and geometric standard deviation to meanlog and sdlog as used in dlnorm
mu <- log(mean_particle_radius_dm)
sdev <- sqrt(log(sd_particle_radius_dm)^2)
proportion_particles <- dlnorm(particle_radius_dm, meanlog = mu, sdlog = sdev)
print("Note: Since logScale==TRUE, the mean is interpreted as the geometric mean and the sd as the geometric standard deviation.")
} else {
proportion_particles <- dnorm(particle_radius_dm, mean=mean_particle_radius_dm, sd = sd_particle_radius_dm)
}
# calculate mass of drug in each bin
pdf_particles <- matrix(0, nrow=nrow(total_deposition), ncol=ncol(total_deposition))
# deposited_particles <- matrix(0, nrow=nrow(total_deposition), ncol=ncol(total_deposition))
for (gen in 1:nrow(total_deposition)) {
pdf_particles[gen,] <- total_deposition[gen,]*proportion_particles
}
## MoBi - see email from Juri Solodenko (AW: Question about dissolution in MoBi) from May 1, 2020
# calculate total volume of all particles, NOTE: volume will be in L
total_volume_ <- 0
add_up_r3 <- 0
for (gen in 1:nrow(pdf_particles)) {
for (radius in 1:ncol(pdf_particles)) {
add_up_r3 <- add_up_r3 + pdf_particles[gen,radius]*(particle_radius_dm[radius]^3)
}
}
total_volume <- (4/3)*pi*add_up_r3
# calculate number of particles factor
number_of_particles_factor <- matrix(0, ncol=ncol(total_deposition))
number_of_particles_factor <- colSums(pdf_particles)/total_volume
distribution_across_gens <- matrix(0, nrow=nrow(total_deposition), ncol=ncol(total_deposition))
for (column in 1:ncol(pdf_particles)) {
distribution_across_gens[,column] <- pdf_particles[,column]/sum(pdf_particles[,column])
}
## Boger
# normalizing_factor <- 0
# for (gen in 1:nrow(pdf_particles)) {
# normalizing_factor <- normalizing_factor + trapz(particle_radius_dm, pdf_particles[gen,])
# }
# pdf_particles <- (pdf_particles / normalizing_factor) * dose
#
# # calculate number of particles in each bin
# mass_of_one <- (particle_radius_dm^3)*4*pi/3 * drug_density_g
# for (gen in 1:nrow(pdf_particles)) {
# deposited_particles[gen,] <- pdf_particles[gen,] / mass_of_one
# }
output <- list("number_of_particles_factor" = number_of_particles_factor,
#"pdf_particles" = pdf_particles,
"distribution_across_gens" = distribution_across_gens)
return(output)
}