-
Notifications
You must be signed in to change notification settings - Fork 0
/
Copy pathCox regression - Osteopenia subset.R
560 lines (509 loc) · 29 KB
/
Cox regression - Osteopenia subset.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
library(survival)
library(survminer)
library(dplyr)
library(splines)
if (!require("pacman")) install.packages("pacman")
pacman::p_load(pacman, rio)
library(rcompanion)
library(car)
library(skimr)
library(mice)
library(VIM)
library(devtools)
library(MAMI)
library(glmnet)
library(plotmo)
library(Hmisc)
library(rms)
library(pec)
library(riskRegression)
library(mlr)
# Import dataset
MOF <- import("C:/Users/BCS de Vries/Documents/Studie/Afstuderen/Data/Cox_regression_MOF_2020_final_subset_11_02.csv")
head(MOF)
attach(MOF)
View(MOF)
# Factor categorical variables
MOF$Gender <- factor(Gender,
levels = c(0,1),
labels = c("male","female"))
MOF$pos_family_history <- factor(pos_family_history,
levels = c(0,1),
labels = c("no","yes"))
MOF$prior_falls <- factor(prior_falls,
levels = c(0,1),
labels = c("no","yes"))
MOF$bedridden <- factor(bedridden,
levels = c(0,1),
labels = c("no","yes"))
MOF$backpain <- factor(backpain,
levels = c(0,1),
labels = c("no","yes"))
MOF$COCP <- factor(COCP,
levels = c(0,1),
labels = c("no","yes"))
MOF$Breastfeeding <- factor(Breastfeeding,
levels = c(0,1),
labels = c("no","yes"))
MOF$Duizeligheid_DBC <- factor(Duizeligheid_DBC,
levels = c(0,1),
labels = c("no","yes"))
MOF$DM <- factor(DM,
levels = c(0,1),
labels = c("no","yes"))
MOF$CVD <- factor(CVD,
levels = c(0,1),
labels = c("no","yes"))
MOF$IBD <- factor(IBD,
levels = c(0,1),
labels = c("no","yes"))
MOF$CVA <- factor(CVA,
levels = c(0,1),
labels = c("no","yes"))
MOF$Epilepsy <- factor(Epilepsy,
levels = c(0,1),
labels = c("no","yes"))
MOF$Systemic_auto_immune <- factor(Systemic_auto_immune,
levels = c(0,1),
labels = c("no","yes"))
MOF$rheuma_art <- factor(rheuma_art,
levels = c(0,1),
labels = c("no","yes"))
MOF$malabsorption <- factor(malabsorption,
levels = c(0,1),
labels = c("no","yes"))
MOF$renal_insufficiency <- factor(renal_insufficiency,
levels = c(0,1),
labels = c("no","yes"))
MOF$corticosteroids <- factor(corticosteroids,
levels = c(0,1),
labels = c("no","yes"))
MOF$Collaps_DBC <- factor(Collaps_DBC,
levels = c(0,1),
labels = c("no","yes"))
MOF$Delier_dementie_DBC <- factor(Delier_dementie_DBC,
levels = c(0,1),
labels = c("no","yes"))
MOF$weigth_under_67 <- factor(weigth_under_67,
levels = c(0,1),
labels = c("no","yes"))
MOF$weight_under_60 <- factor(weight_under_60,
levels = c(0,1),
labels = c("no","yes"))
MOF$Change_length <- factor(Change_length,
levels = c(0,1),
labels = c("no","yes"))
MOF$More_6_coffee <- factor(More_6_coffee,
levels = c(0,1),
labels = c("no","yes"))
MOF$Sun_exposure <- factor(Sun_exposure,
levels = c(0,1),
labels = c("no","yes"))
MOF$Fat_fish_diet <- factor(Fat_fish_diet,
levels = c(0,1),
labels = c("no","yes"))
MOF$veg_diet <- factor(veg_diet,
levels = c(0,1),
labels = c("no","yes"))
MOF$Vitam_pills <- factor(Vitam_pills,
levels = c(0,1),
labels = c("no","yes"))
MOF$Daily_margarin <- factor(Daily_margarin,
levels = c(0,1),
labels = c("no","yes"))
MOF$Lab_GFR_decreased <- factor(Lab_GFR_decreased,
levels = c(0,1),
labels = c("GFR>=90", "GFR<90"))
#add Hazard function
hazard <- nelsonaalen(MOF, time_w, MOF_fracture)
hazard
#Delete variable rec_id, add hazard
ncol(MOF)
MOF_impute <- MOF[2:50]
View(MOF_impute)
MOF_impute <- cbind(MOF_impute, hazard) # add hazard
View(MOF_impute)
# Add interaction terms for imputation with MICE
# Interaction Age (68.53) with hours per week active (25.87), mean in ().
Meth <- make.method(MOF_impute)
expr_age.hours <- expression((Age - 68.53) * (Hours_active - 25.87))
MOF_impute$Age.hours <- with(MOF_impute, eval(expr_age.hours))
Meth["Age.hours"] <- paste("~I(", expr_age.hours, ")", sep = "")
# Interaction Age (68.53) with T_score_HIP (-1.45), mean in ().
expr_age.T_score_HIP <- expression((Age - 68.53) * (T_score_HIP - (-1.45)))
MOF_impute$Age.T_score_HIP <- with(MOF_impute, eval(expr_age.T_score_HIP))
Meth["Age.T_score_HIP"] <- paste("~I(", expr_age.T_score_HIP, ")", sep = "")
# Interaction Age (68.53) with T_score_LWK (-1.50), mean in ().
expr_age.T_score_LWK <- expression((Age - 68.53) * (T_score_LWK - (-1.50)))
MOF_impute$Age.T_score_LWK <- with(MOF_impute, eval(expr_age.T_score_LWK))
Meth["Age.T_score_LWK"] <- paste("~I(", expr_age.T_score_LWK, ")", sep = "")
# Interaction Age (68.53) with Gender, gemiddelde in ().
expr_age.Gender <- expression((Age - 68.53) * Gender_interaction)
MOF_impute$Age.Gender <- with(MOF_impute, eval(expr_age.Gender))
Meth["Age.Gender"] <- paste("~I(", expr_age.Gender, ")", sep = "")
# Interaction Gender with hours per week active (25.87), mean in ().
expr_Gender.hours <- expression((Gender_interaction) * (Hours_active - 25.87))
MOF_impute$Gender.hours <- with(MOF_impute, eval(expr_Gender.hours))
Meth["Gender.hours"] <- paste("~I(", expr_Gender.hours, ")", sep = "")
# Interaction Gender with T_score_HIP (-1.45), mean in ().
expr_Gender.T_score_HIP <- expression((Gender_interaction) * (T_score_HIP - (-1.45)))
MOF_impute$Gender.T_score_HIP <- with(MOF_impute, eval(expr_Gender.T_score_HIP))
Meth["Gender.T_score_HIP"] <- paste("~I(", expr_Gender.T_score_HIP, ")", sep = "")
# Interaction Gender with T_score_LWK (-1.50), mean in ().
expr_Gender.T_score_LWK <- expression((Gender_interaction) * (T_score_LWK - (-1.50)))
MOF_impute$Gender.T_score_LWK <- with(MOF_impute, eval(expr_Gender.T_score_LWK))
Meth["Gender.T_score_LWK"] <- paste("~I(", expr_Gender.T_score_LWK, ")", sep = "")
#Prepare prediction matrix of MICE
Pred <- quickpred(MOF_impute, include = c("hazard", "Gender", "Age","MOF_fracture"),
exclude = c( "time_m","time_w"),
minpuc = 0.5, mincor = 0.1)
Pred[c("duration_menopause", "Lab_GFR_from_90","children","Hours_active"), "Age.hours"] <- 0
Pred[c("duration_menopause", "Lab_GFR_from_90","children","T_score_HIP"), "Age.T_score_HIP"] <- 0
Pred[c("duration_menopause", "Lab_GFR_from_90","children","T_score_LWK"), "Age.T_score_LWK"] <- 0
Pred[c("duration_menopause", "Lab_GFR_from_90","children","Gender"), "Age.Gender"] <- 0
Pred[c("duration_menopause", "Lab_GFR_from_90","children","Hours_active"), "Gender.hours"] <- 0
Pred[c("duration_menopause", "Lab_GFR_from_90","children","T_score_HIP"), "Gender.T_score_HIP"] <- 0
Pred[c("duration_menopause", "Lab_GFR_from_90","children","T_score_LWK"), "Gender.T_score_LWK"] <- 0
length(MOF_impute)
Pred[1:57, "Gender_interaction"] <- 0
# mean number of predictors
rowSums(Pred)
select <- which(rowSums(Pred)>0)
View(select)
select_no_interaction <- select[1:29]
mean(select_no_interaction)
#Impute with conditions and correct predictormatrix
MOF_imputed_good <- mice(MOF_impute, pred = Pred, meth = Meth,
m=30, maxit = 50, seed=2795, print = F)
MOF_imputed_good
densityplot(MOF_imputed_good)
# complete datasets seperately
complete1 <- complete(MOF_imputed_good,1)
complete2 <- complete(MOF_imputed_good,2)
complete3 <- complete(MOF_imputed_good,3)
complete4 <- complete(MOF_imputed_good,4)
complete5 <- complete(MOF_imputed_good,5)
complete6 <- complete(MOF_imputed_good,6)
complete7 <- complete(MOF_imputed_good,7)
complete8 <- complete(MOF_imputed_good,8)
complete9 <- complete(MOF_imputed_good,9)
complete10 <- complete(MOF_imputed_good,10)
complete11 <- complete(MOF_imputed_good,11)
complete12 <- complete(MOF_imputed_good,12)
complete13 <- complete(MOF_imputed_good,13)
complete14 <- complete(MOF_imputed_good,14)
complete15 <- complete(MOF_imputed_good,15)
complete16 <- complete(MOF_imputed_good,16)
complete17 <- complete(MOF_imputed_good,17)
complete18 <- complete(MOF_imputed_good,18)
complete19 <- complete(MOF_imputed_good,19)
complete20 <- complete(MOF_imputed_good,20)
complete21 <- complete(MOF_imputed_good,21)
complete22 <- complete(MOF_imputed_good,22)
complete23 <- complete(MOF_imputed_good,23)
complete24 <- complete(MOF_imputed_good,24)
complete25 <- complete(MOF_imputed_good,25)
complete26 <- complete(MOF_imputed_good,26)
complete27 <- complete(MOF_imputed_good,27)
complete28 <- complete(MOF_imputed_good,28)
complete29 <- complete(MOF_imputed_good,29)
complete30 <- complete(MOF_imputed_good,30)
# complete a stacked dataset with weight 1/w.
MOF_stacked <- complete1
for (i in 2:30){
MOF_stacked <- rbind(MOF_stacked, complete(MOF_imputed_good, i))}
MOF_stacked$w <- 1/30
#check if there are no values imputed which are not possible
min(MOF_stacked$Lab_bse)
min(MOF_stacked$Lab_calcium)
min(MOF_stacked$Lab_albumin)
min(MOF_stacked$Lab_TSH)
min(MOF_stacked$Lab_vit_d)
min(MOF_stacked$Lab_GFR_from_90)
min(MOF_stacked$Hours_active)
min(MOF_stacked$Total_daily_calcium)
min(MOF_stacked$children)
# Check assumption of proportional hazard for complete model
cox_function_pha <- coxph(Surv(time_w,MOF_fracture) ~ Age + Gender + prior_falls + pos_family_history + bedridden + backpain + children + COCP + Breastfeeding + ns(duration_menopause, 3) + DM + CVD + IBD + CVA + Epilepsy + Systemic_auto_immune + rheuma_art + malabsorption + renal_insufficiency + corticosteroids + Collaps_DBC + Delier_dementie_DBC + Duizeligheid_DBC + weigth_under_67 + weight_under_60 + Change_length + Hours_active + Total_daily_calcium + More_6_coffee + Sun_exposure + Fat_fish_diet + veg_diet + Vitam_pills + Daily_margarin + Lab_bse + ns(Lab_calcium, 4) + Lab_albumin + Lab_TSH + ns(Lab_vit_d, 3) + Lab_GFR_decreased + ns(Lab_GFR_from_90,3) + T_score_LWK + T_score_HIP + Weight + Length + Hours_active:Age + T_score_HIP:Age + T_score_LWK:Age + Gender:Age + Hours_active:Gender + T_score_HIP:Gender + T_score_LWK:Gender,
data = MOF_stacked,
weights = w,
x = TRUE,
y = TRUE)
summary(cox_function_pha)
print(cox.zph(cox_function_pha))
# Check assumption of proportional hazard for final model
cox_function_final_model <- coxph(Surv(time_w, MOF_fracture)~ Age + Gender + prior_falls + ns(duration_menopause,3) + T_score_HIP + Epilepsy + T_score_HIP:Age,
data = MOF_stacked,
weights = w,
x = TRUE,
y = TRUE)
print(cox.zph(cox_function_final_model))
# Model for LASSO and fit with glmnet
cox_function_all <- coxph(Surv(time_w, MOF_fracture) ~ Age + Gender + prior_falls + pos_family_history + bedridden + backpain + children + COCP + Breastfeeding + ns(duration_menopause, 3) + DM + CVD + IBD + CVA + Epilepsy + Systemic_auto_immune + rheuma_art + malabsorption + renal_insufficiency + corticosteroids + Collaps_DBC + Delier_dementie_DBC + Duizeligheid_DBC + weigth_under_67 + weight_under_60 + Change_length + Hours_active + Total_daily_calcium + More_6_coffee + Sun_exposure + Fat_fish_diet + veg_diet + Vitam_pills + Daily_margarin + Lab_bse + ns(Lab_calcium, 4) + Lab_albumin + Lab_TSH + ns(Lab_vit_d, 3) + Lab_GFR_decreased + ns(Lab_GFR_from_90,3) + T_score_LWK + T_score_HIP + Weight + Length + Hours_active:Age + T_score_HIP:Age + T_score_LWK:Age + Gender:Age + Hours_active:Gender + T_score_HIP:Gender + T_score_LWK:Gender,
data = MOF_stacked,
weights = w,
x = TRUE,
y = TRUE)
fit = glmnet(
x = cox_function_all$x,
y = cox_function_all$y,
alpha = 1, family = "cox")
# Compute Coxph with LASSO on every dataset to get mean penalty factor lambda (both 1 SE and min)
coef_min_total <- 0
coef_1se_total <- 0
for (i in 1:30){
data1 <- as.data.frame(complete(MOF_imputed_good, i))
cox_function_LASSO <- coxph(Surv(time_w,MOF_fracture) ~ Age + Gender + prior_falls + pos_family_history + bedridden + backpain + children + COCP + Breastfeeding + ns(duration_menopause, 3) + DM + CVD + IBD + CVA + Epilepsy + Systemic_auto_immune + rheuma_art + malabsorption + renal_insufficiency + corticosteroids + Collaps_DBC + Delier_dementie_DBC + Duizeligheid_DBC + weigth_under_67 + weight_under_60 + Change_length + Hours_active + Total_daily_calcium + More_6_coffee + Sun_exposure + Fat_fish_diet + veg_diet + Vitam_pills + Daily_margarin + Lab_bse + ns(Lab_calcium, 4) + Lab_albumin + Lab_TSH + ns(Lab_vit_d, 3) + Lab_GFR_decreased + ns(Lab_GFR_from_90,3) + T_score_LWK + T_score_HIP + Weight + Length + Hours_active:Age + T_score_HIP:Age + T_score_LWK:Age + Gender:Age + Hours_active:Gender + T_score_HIP:Gender + T_score_LWK:Gender,
data = data1,
x = TRUE,
y = TRUE)
coxlasso1 <- glmnet(
x = cox_function_LASSO$x,
y = cox_function_LASSO$y,
alpha=1, family = 'cox')
cvfit1 <- cv.glmnet(
x = cox_function_LASSO$x,
y = cox_function_LASSO$y, family = "cox")
coef_1se_total <- c(coef_1se_total, cvfit1$lambda.1se)
coef_min_total <- c(coef_min_total, cvfit1$lambda.min)
}
coef_1se_total <- coef_1se_total[2:31]
mean_LASSO_1se <- mean(coef_1se_total)
plot(cvfit1)
coef_select_1se <- coef(fit, s=mean_LASSO_1se)
select_1se <- which(or(coef_select_1se>0,coef_select_1se<0))
coef_select_1se[select_1se, 1]
coef_min_total <- coef_min_total[2:31]
mean_LASSO_min <- mean(coef_min_total)
coef_select_min <- coef(fit, s=mean_LASSO_min)
select_min <- which(or(coef_select_min>0,coef_select_min<0))
coef_select_min[select_min, 1]
#Fit final Cox model on all 30 datasets and pool for lambda 1se
fit.Cox <- with(data=MOF_imputed_good, exp=coxph(Surv(time_w,MOF_fracture) ~ Age))
Cox.pool <- summary(pool(fit.Cox))
Cox.pool
pool.HR <- exp(cbind(Cox.pool[,1], (Cox.pool[,1]-1.96*(Cox.pool[,2])),
(Cox.pool[,1]+1.96*(Cox.pool[,2]))))
colnames(pool.HR) <- (c("HR", "95% LO", "95% UP"))
pool.HR
outcome_Cox <- cbind(Cox.pool, pool.HR)
outcome_Cox <- round(outcome_Cox,3)
print(outcome_Cox)
#Fit final Cox model on all 30 datasets and pool for lambda min
fit.Cox_min <- with(data=MOF_imputed_good, exp=coxph(Surv(time_w,MOF_fracture) ~ Age + Gender + prior_falls + CVD + Delier_dementie_DBC + Change_length + Hours_active + Fat_fish_diet + Lab_GFR_decreased + ns(Lab_GFR_from_90, 3) + Age:T_score_HIP + T_score_HIP))
Cox.pool_min <- summary(pool(fit.Cox_min))
Cox.pool_min
pool.HR_min <- exp(cbind(Cox.pool_min[,1], (Cox.pool_min[,1]-1.96*(Cox.pool_min[,2])),
(Cox.pool_min[,1]+1.96*(Cox.pool_min[,2]))))
colnames(pool.HR_min) <- (c("HR", "95% LO", "95% UP"))
pool.HR_min
outcome_Cox_min <- cbind(Cox.pool_min, pool.HR_min)
outcome_Cox_min <- round(outcome_Cox_min,3)
print(outcome_Cox_min)
# Cross validation loop for 1se
matrix_c_index_total <- as.data.frame(matrix(c(1,2,3,4,5,6,7,8,9,10)))
c_index <- as.data.frame(matrix(c("training", "test" )))
for (i in 1:30){
data5 <- as.data.frame(complete(MOF_imputed_good, i))
dd <- datadist(data5)
options(datadist="dd")
cv_cox_function <- cph(Surv(time_w,MOF_fracture) ~ Age,
data = data5,
surv = TRUE,
x = TRUE,
y = TRUE)
cv_cox_function <- pec::cindex(cv_cox_function,
formula=Surv(time_w,MOF_fracture) ~ .,
data = data5,
splitMethod = "bootcv",
B = 10,
confInt = TRUE,
keep.matrix = TRUE,)
matrix_c_index <- as.data.frame(cv_cox_function$BootstrapCrossValCindexMat)
matrix_c_index_total <- cbind(matrix_c_index_total, matrix_c_index)
c_index_data2 <- as.data.frame(matrix(c(cv_cox_function$AppCindex, cv_cox_function$BootCvCindex)))
c_index <- cbind(c_index, c_index_data2)
}
colnames(c_index) <- c("test/training", "dataframe 1", "dataframe 2", "dataframe3", "dataframe4", "dataframe5", "dataframe6", "dataframe7", "dataframe8", "dataframe9", "dataframe10", "dataframe11", "dataframe12", "dataframe13", "dataframe14", "dataframe15", "dataframe16", "dataframe17", "dataframe18", "dataframe19", "dataframe20", "dataframe21", "dataframe22", "dataframe23", "dataframe24", "dataframe25", "dataframe26", "dataframe27", "dataframe28", "dataframe29", "dataframe30")
View(c_index)
c_index_adj <- c_index[,2:31]
c_index_matrix <- data.matrix(c_index_adj)
c_index_mean <- rowMeans(c_index_matrix)
training <- "concordance index training"
test <- "concordance index test"
cat("concordance index training", c_index_mean[1],"\n",
"concordance index test", c_index_mean[2])
# confidence interval using Rubin's Rules
colnames(matrix_c_index_total) <- c("#iteration", "dataframe 1", "dataframe 2", "dataframe3", "dataframe4", "dataframe5", "dataframe6", "dataframe7", "dataframe8", "dataframe9", "dataframe10", "dataframe11", "dataframe12", "dataframe13", "dataframe14", "dataframe15", "dataframe16", "dataframe17", "dataframe18", "dataframe19", "dataframe20", "dataframe21", "dataframe22", "dataframe23", "dataframe24", "dataframe25", "dataframe26", "dataframe27", "dataframe28", "dataframe29", "dataframe30")
mean <- 0
variance <- 0
for (i in 2:31){
mean[i] <- mean(matrix_c_index_total[,i])
variance[i] <- var(matrix_c_index_total[,i])
}
mean_good <- mean[2:31]
variance_good <- variance[2:31]
pooled<- pool.scalar(Q=mean_good, U=variance_good, n=1770)
pooled$qbar
Confidence_interval_low = pooled$qbar-1.96*sqrt(pooled$t)
Confidence_interval_high = pooled$qbar+1.96*sqrt(pooled$t)
CI <- c(Confidence_interval_low,Confidence_interval_high)
print(CI)
# Cross validation loop for lambda min
matrix_c_index_total_min <- as.data.frame(matrix(c(1,2,3,4,5,6,7,8,9,10)))
c_index_min <- as.data.frame(matrix(c("training", "test" )))
for (i in 1:30){
data6 <- as.data.frame(complete(MOF_imputed_good, i))
dd <- datadist(data6)
options(datadist="dd")
cv_cox_function_min <- cph(Surv(time_w,MOF_fracture) ~ Age + Gender + prior_falls + CVD + Delier_dementie_DBC + Change_length + Hours_active + Fat_fish_diet + Lab_GFR_decreased + rcs(Lab_GFR_from_90, 3) + Age*T_score_HIP + T_score_HIP,
data = data6,
surv = TRUE,
x = TRUE,
y = TRUE)
cv_cox_function_min <- pec::cindex(cv_cox_function_min,
formula=Surv(time_w,MOF_fracture) ~ .,
data = data6,
splitMethod = "bootcv",
B = 10,
confInt = TRUE,
keep.matrix = TRUE,)
matrix_c_index_min <- as.data.frame(cv_cox_function_min$BootstrapCrossValCindexMat)
matrix_c_index_total_min <- cbind(matrix_c_index_total_min, matrix_c_index_min)
c_index_data3_min <- as.data.frame(matrix(c(cv_cox_function_min$AppCindex, cv_cox_function_min$BootCvCindex)))
c_index_min <- cbind(c_index_min, c_index_data3_min)
}
colnames(c_index_min) <- c("test/training", "dataframe 1", "dataframe 2", "dataframe3", "dataframe4", "dataframe5", "dataframe6", "dataframe7", "dataframe8", "dataframe9", "dataframe10", "dataframe11", "dataframe12", "dataframe13", "dataframe14", "dataframe15", "dataframe16", "dataframe17", "dataframe18", "dataframe19", "dataframe20", "dataframe21", "dataframe22", "dataframe23", "dataframe24", "dataframe25", "dataframe26", "dataframe27", "dataframe28", "dataframe29", "dataframe30")
View(c_index_min)
colnames(matrix_c_index_total_min) <- c("#iteration", "dataframe 1", "dataframe 2", "dataframe3", "dataframe4", "dataframe5", "dataframe6", "dataframe7", "dataframe8", "dataframe9", "dataframe10", "dataframe11", "dataframe12", "dataframe13", "dataframe14", "dataframe15", "dataframe16", "dataframe17", "dataframe18", "dataframe19", "dataframe20", "dataframe21", "dataframe22", "dataframe23", "dataframe24", "dataframe25", "dataframe26", "dataframe27", "dataframe28", "dataframe29", "dataframe30")
View(matrix_c_index_total_min)
c_index_adj_min <- c_index_min[,2:31]
c_index_matrix_min <- data.matrix(c_index_adj_min)
c_index_mean_min <- rowMeans(c_index_matrix_min)
cat("concordance index training", c_index_mean_min[1],"\n",
"concordance index test", c_index_mean_min[2])
#confidence interval using Rubin's Rules
mean <- 0
variance <- 0
for (i in 2:31){
mean[i] <- mean(matrix_c_index_total_min[,i])
variance[i] <- var(matrix_c_index_total_min[,i])
}
mean_good <- mean[2:31]
variance_good <- variance[2:31]
pooled<- pool.scalar(Q=mean_good, U=variance_good, n=1770)
pooled$qbar
Confidence_interval_low = pooled$qbar-1.96*sqrt(pooled$t)
Confidence_interval_high = pooled$qbar+1.96*sqrt(pooled$t)
CI <- c(Confidence_interval_low,Confidence_interval_high)
print(CI)
# Shiny Tool
library(rsconnect)
rsconnect::setAccountInfo(name='majorosteoporoticfracture', token='F15BD5A96C6A91A27B59E1EEE2410AEC', secret='EWSKAkJRir8ieRD2YBGYcRtkSxN2UMOuwvc2b/6e')
library(shiny)
library(scales)
ui <- fluidPage(
titlePanel("Major Osteoporotic Fracture Risk Calculator"),
wellPanel(fluidRow(
column(6, numericInput("age", "Age:", 50, min = 50, max = 110)),
column(6, selectInput("gender", "Gender:", c("male","female"), selected = "female", multiple=FALSE)),
),
fluidRow(
column(6, selectInput("prior_falls", "Prior Falls:", c("no","yes"), selected="no")),
column(6, selectInput("cvd", "Cardiovascular disease:", c("no","yes"), selected="no")),
),
fluidRow(
column(6, selectInput("delier_dementie", "Delirium or dementia:", c("no","yes"), selected="no")),
column(6, selectInput("change_length", "Change in length in recent years:", c("no","yes"), selected="no")),
),
fluidRow(
column(6, numericInput("hours_active", "Hours of moderate activity per week:", 0, min=0, max=168)),
column(6, selectInput("fat_fish", "Dietary use of fat fish (at least twice a week):", c("no","yes"), selected="no")),
),
fluidRow(
column(6, numericInput("eGFR", "Last known eGFR: (if > 90 fill in 90)", 90, min=0, max=90)),
column(6, numericInput("T_score_hip", "T-score of the hip:",-1.5, min=-5, max = 5)),
),
fluidRow(
column(3, actionButton("go", "Predict risk"))
)),
fluidRow(
column(12, verbatimTextOutput("text"))
),
wellPanel(
verbatimTextOutput("text_age"),
)
)
server <- function(input, output) {
newdata <- data.frame(Age=70, Gender = "female", prior_falls = "no", CVD = "yes", Delier_dementie_DBC = "no", Change_length="yes", Hours_active = 7, Fat_fish_diet = "yes", Lab_GFR_decreased = "GFR<90", Lab_GFR_from_90 = 1, T_score_HIP = -1.5)
data <- eventReactive(input$go, {
newdata$Gender <- factor(newdata$Gender,
levels = c(0,1),
labels = c("male","female"))
newdata$prior_falls <- factor(newdata$prior_falls,
levels = c(0,1),
labels = c("no","yes"))
newdata$Fat_fish_diet <- factor(newdata$Fat_fish_diet,
levels = c(0,1),
labels = c("no","yes"))
newdata$CVD <- factor(newdata$CVD,
levels = c(0,1),
labels = c("no","yes"))
newdata$Delier_dementie_DBC <- factor(newdata$Delier_dementie_DBC,
levels = c(0,1),
labels = c("no","yes"))
newdata$Change_length <- factor(newdata$Change_length,
levels = c(0,1),
labels = c("no","yes"))
newdata$Lab_GFR_decreased <- factor(newdata$Lab_GFR_decreased,
levels = c(0,1),
labels = c("GFR>=90", "GFR<90"))
converted_eGFR = 90-input$eGFR
# eGFR
if (converted_eGFR > 0 ){
nierfunctie = "GFR<90"
} else {
nierfunctie = "GFR>=90"
}
# Prediction on new data
newdata <- data.frame(Age=input$age, Gender = input$gender, prior_falls = input$prior_falls, CVD = input$cvd, Delier_dementie_DBC = input$delier_dementie, Change_length=input$change_length, Hours_active = input$hours_active, Fat_fish_diet = input$fat_fish, Lab_GFR_decreased = nierfunctie, Lab_GFR_from_90 = converted_eGFR, T_score_HIP = input$T_score_hip)
pred_survival_matrix <- as.data.frame(matrix(c("3 year", "5 year")))
for (i in 1:30){
data10 <- as.data.frame(complete(MOF_imputed_good, i))
dd <- datadist(data10)
options(datadist="dd")
cox_function_final_model <- coxph(Surv(time_w, MOF_fracture) ~ Age + Gender + prior_falls + CVD + Delier_dementie_DBC + Change_length + Hours_active + Fat_fish_diet + Lab_GFR_decreased + ns(Lab_GFR_from_90, 3) + Age:T_score_HIP + T_score_HIP,
data = data10,
x = TRUE,
y = TRUE)
fit <- predictSurvProb(cox_function_final_model, newdata = newdata, times = c(156,260))
fit_dataframe<- as.data.frame(matrix(fit, nrow=2, ncol=1))
pred_survival_matrix <- cbind(pred_survival_matrix, fit_dataframe)
}
prediction_all <- pred_survival_matrix[2:31]
Three_year_risk = (1-rowMeans(prediction_all[1,]))
Five_year_risk = (1-rowMeans(prediction_all[2,]))
Three_year_risk_text <- paste("The 3 year risk of Major Osteoporotic Fracture =", percent(Three_year_risk))
Five_year_risk_text <- paste("The 5 year risk of Major Osteoporotic Fracture =", percent(Five_year_risk))
paste(Three_year_risk_text, Five_year_risk_text, "" , "With Major Osteoporotic Fracture defined as a fracture from the hip, wrist, spine or humerus", sep="\n")
})
output$text <- renderText({
data()
})
output$text_age <- renderText({paste("Disclaimer: The Major Osteoporotic Fracture Risk Calculator still needs further (external) validation, and clinical evaluation before it can support decision making.",
"",
"Risk factors should be used as follows:",
"",
"Age: Current age in years",
"Gender: Male or female",
"Prior falls: if prior falling occured, tick yes",
"Cardiovascular disease: a history or a current comorbidity related to the cardiovascular system. Examples are arrythmia, coronary heart disease and peripherial arterial disease amongst others.",
"Delirium or dementia: a history of delirium or presence of dementia.",
"Change in length in recent years: fill in yes when the patients noticed diminishing length in the last few years.",
"Hours of moderate activity per week: moderate activity as defined by the WHO. Examples include gardening, housework and walking domestic animals.",
"Dietary use of fat fish: tick yes if a patient reports including fat fish at least twice a week in his/her diet.",
"Last known eGFR: Enter the last known eGFR with no decimal. If the eGFR exceeds 90, fill in 90.",
"T-score of the hip: Enter the last known T-score of the hip using a comma.",
sep="\n")})
}
shinyApp(ui = ui, server = server)