-
Notifications
You must be signed in to change notification settings - Fork 0
/
Script_calling_weight_trajectory.R
813 lines (696 loc) · 39.5 KB
/
Script_calling_weight_trajectory.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
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
#R script for calling stable weight, weight gain, weight loss, or weight cycle
############### stable weight trajectory ##################
############################################################
#get the the maximum and minimum weight per individual first
library(dplyr)
grp<-group_by(your_data,individual_ID)
max_annualKG <-aggregate(annualKG ~ individual_ID, data = your_data, max)
min_annualKG <-aggregate(annualKG ~ individual_ID, data = your_data, min)
#only those whose maximum weight change is <5% from the baseline are classified as having a stable weight
your_data$perc_maxfirst_annualKG <- your_data$max_annualKG/your_data$first_annualKG*100-100
your_data$perc_minfirst_annualKG <- your_data$min_annualKG/your_data$first_annualKG*100-100
#if 5% is used as the cutoff for clinically meaningful weight change
your_data$stable_weight_5 <-ifelse((your_data$perc_maxfirst_annualKG < 5) & (your_data$perc_minfirst_annualKG >-5),1,0)
#if 10% is used as the cutoff for clinically meaningful weight change
your_data$stable_weight_10 <-ifelse((your_data$perc_maxfirst_annualKG < 10) & (your_data$perc_minfirst_annualKG >-10),1,0)
############################################################
################# Weight loss trajectory ##################
############################################################
#Three criteria to define a weight loss
#either 5% or 10% as the cutoff for clinically meaningful weight change
#(1) the net weight loss from the first to the last annual weight over the measured period was > 0;
#computing the first and last annual weight
library(data.table)
DT <- as.data.table(your_data)
tmp1<-DT[, .SD[1], by=individual_ID]
tmp2<-DT[, .SD[.N], by=individual_ID]
#SD subset of dataset
#.N is the number of rows in each group
tmp1 <- as.data.frame(tmp1)
tmp2 <- as.data.frame(tmp2)
tmp1$first_annualKG <-tmp1$annualKG
tmp2$last_annualKG <- tmp2$annualKG
#compute the net change from the first to the last annual weight (in percentage)
your_data$perc_lastfirst_annualKG <-your_data$last_annualKG/your_data$first_annualKG*100-100
#(2) the maximum weight loss from baseline was ≥ 5%;
#(3) overall the individual had more weight loss than weight gain over time and this was quantified by
#(a) the maximum weight gain from baseline was < 5%,
#or (b) the amount of maximum weight gain from baseline was < 45% of the overall weight change magnitude (maximum - minimum)
#The cutoff of 45% was chosen based on manual visual inspection of the weight trajectory plots among a random group of 100 individuals.
#You can change 45% to other values based on how weight trajectory looks like in your Biobank/cohort
#compute the difference between the maximum and the first annual weights per individual
your_data$diff_maxfirst_annualKG <- your_data$max_annualKG - your_data$first_annualKG
#compute the difference between the maximum and minimum weights per individual
your_data$diff_maxmin_annualKG <- your_data$max_annualKG - your_data$min_annualKG
#creating the weight loss trajectory binary variable based on the 3 criteria using either 5% or 10% cutoff
your_data$weight_loss_5 <- ifelse((your_data$perc_lastfirst_annualKG < 0 & your_data$perc_minfirst_annualKG <= -5 & (your_data$perc_maxfirst_annualKG < 5 | (abs(your_data$diff_maxfirst_annualKG) < 0.45*your_data$diff_maxmin_annualKG)) ),1,0)
your_data$weight_loss_10 <- ifelse((your_data$perc_lastfirst_annualKG < 0 & your_data$perc_minfirst_annualKG <= -10 & (your_data$perc_maxfirst_annualKG < 5 | (abs(your_data$diff_maxfirst_annualKG) < 0.45*your_data$diff_maxmin_annualKG)) ),1,0)
############################################################
################# Weight gain trajectory ##################
############################################################
#Three criteria to define a weight gain
#either 5% or 10% as the cutoff for clinically meaningful weight change
#(1) the net weight gain from the first to the last annual weight over the measured period was > 0;
#(2) the maximum weight gain from baseline was ≥ 5%; ;
#(3) overall the individual had more weight gain than weight loss over time and this was quantified by
#(a) the maximum weight loss from baseline was < 5%,
#or (b) he amount of maximum weight loss from baseline was < 45% of the overall weight change magnitude (maximum - minimum),
#The cutoff of 45% was chosen based on manual visual inspection of the weight trajectory plots among a random group of 100 individuals.
#You can change 45% to other values based on how weight trajectory looks like in your Biobank/cohort
#compute the difference between the maximum and the first annual weights per individual
your_data$diff_minfirst_annualKG <- your_data$min_annualKG - your_data$first_annualKG
#creating the weight gain trajectory binary variable based on the 3 criteria using either 5% or 10% cutoff
your_data$weight_gain_5 <- ifelse((your_data$perc_lastfirst_annualKG > 0 & your_data$perc_maxfirst_annualKG >= 5 & (your_data$perc_minfirst_annualKG > -5 | (abs(your_data$diff_minfirst_annualKG) < 0.45*your_data$diff_maxmin_annualKG)) ),1,0)
your_data$weight_gain_10 <- ifelse((your_data$perc_lastfirst_annualKG > 0 & your_data$perc_maxfirst_annualKG >= 10 & (your_data$perc_minfirst_annualKG > -5 | (abs(your_data$diff_minfirst_annualKG) < 0.45*your_data$diff_maxmin_annualKG)) ),1,0)
############################################################
################# Weight cycle trajectory #################
############################################################
#weight cycle was identified through either the (1) local minimum/maximum (2) global minimum/maximum approaches
##### (1) local minimum/maximum #####
#first, weight changes between each adjacent annual weights need to be computed (i.e., the weight change from ith annual weight to (i+1)th annual weight)
test<- your_data
for (i in 1:dim(your_data)[1]) {
if (i>1){
#if it is the same individual, compute the year difference, weight different (in kg or %), and weight change per year between the two adjacent annual weights
if (test[i,1]==test[i+1,1]) {
test$year_diff[i+1] <- test$YOMeasure[i+1] - test$YOMeasure[i]
test$annualKG_diff[i+1] <- test$annualKG[i+1] - test$annualKG[i]
test$wt_chng_peryear[i+1] <- test$annualKG_diff[i+1]/test$year_diff[i+1]
test$percdiff_annual[i+1] <- test$annualKG[i+1]/test$annualKG[i]*100-100
}
#if it is not the same individual (i.e., change from the last annual weight for jth individual to the first annual weight for the (j+1)th individual), leave the first row as NA for (j+1)th individual, as it has no preceding weight from the same individual to calculate the weight difference
else if (test[i,1]!=test[i+1,1]) {
test$year_diff[i+1] <- NA
test$annualKG_diff[i+1] <- NA
test$wt_chng_peryear[i+1] <- NA
test$percdiff_annual[i+1] <- NA
}
}
#for the first annual weight of the first individual in the dataset, leave the first row as NA, and compute the weight difference for the second annual weight (compared to the first annual weight)
if (i==1){
test$year_diff[i] <- NA
test$annualKG_diff[i] <- NA
test$wt_chng_peryear[i] <- NA
test$percdiff_annual[i] <- NA
test$year_diff[i+1] <- test$YOMeasure[i+1] - test$YOMeasure[i]
test$annualKG_diff[i+1] <- test$annualKG[i+1] - test$annualKG[i]
test$wt_chng_peryear[i+1] <- test$annualKG_diff[i+1]/test$year_diff[i+1]
test$percdiff_annual[i+1] <- test$annualKG[i+1]/test$annualKG[i]*100-100
}
}
your_data <- test
#second, restrict the weight dataset to the first annual weight, last annual weight, and any annual weight(s) in between that is a inflection point (aka, local peak, could be either local minimum or local maximum)
num_individual <- dim(your_data)[1]
num_column <- dim(your_data)[2]
your_data_subset1 <- data.frame(matrix(NA, nrow = num_individual, ncol = num_column))
for (i in 1:dim(your_data)[1]) {
#if the weight change value (%) between annual weight is not NA or missing
if ((!is.na(your_data$percdiff_annual[i])) & (!is.na(your_data$percdiff_annual[i+1]))) {
#keep the local maximum point
if ((your_data$percdiff_annual[i] > 0) & (your_data$percdiff_annual[i+1] <= 0)) {
your_data_subset1[i,] <- your_data[i,]
}
#keep the local minimum point
else if ((your_data$percdiff_annual[i] < 0) & (your_data$percdiff_annual[i+1] >= 0)) {
your_data_subset1[i,] <- your_data[i,]
}
#otherwise, set this row as NA
else {your_data_subset1[i,] <- your_data_subset1[i,]}
}
#keep the first annual weight
else if (is.na(your_data$percdiff_annual[i])) {
your_data_subset1[i,] <- your_data[i,]
}
#keep the last annual weight
else if (is.na(your_data$percdiff_annual[i+1])){
your_data_subset1[i,] <- your_data[i,]
}
}
your_data_subset1 <- your_data_subset1[which(!is.na(your_data_subset1$annualKG)),]
#using the code above, only the first, local maximum/minimum, and last annual weights per individual were kept
#the rows with NA were removed
#Third, calculating the weight percentage difference between the inflection points (annual weight at inflection point 2 - annual weight at inflection point 1)
#create the % wt change between each turning point
test<-your_data_subset1
for (i in 1:dim(test)[1]) {
if (i>1){
#if it is the same individual, compute the weight different (in %) between the two inflection points
if (test[i,1]==test[i+1,1]){
test$percdiff_turnpt[i+1] <- test$annualKG[i+1]/test$annualKG[i]*100-100
}
#if it is not the same individual (i.e., change from the last annual weight for jth individual to the first annual weight for the (j+1)th individual), leave the first row as NA for (j+1)th individual, as it has no preceding weight from the same individual to calculate the weight difference
else if (test[i,1]!=test[i+1,1]) {
test$percdiff_turnpt[i+1] <- NA
}
}
#for the first annual weight of the first individual in the dataset, leave the first row as NA, and compute the weight difference for the first inflection point (compared to the first annual weight)
if (i==1){
test$percdiff_turnpt[i] <- NA
test$percdiff_turnpt[i+1] <- test$annualKG[i+1]/test$annualKG[i]*100-100
}
}
your_data_subset1 <-test
#Fourth (optional), only if there is any plateau points
#using the codes above, there could be some plateau points kept in the dataset, which we also need to remove (aka, for one inflection point, the annual weight right after it has the same weight value)
num_individual_subset1 <- dim(your_data_subset1)[1]
num_column_subset1 <- dim(your_data_subset1)[2]
your_data_subset2 <- data.frame(matrix(NA, nrow = num_individual_subset1, ncol = num_column_subset1))
for (i in 1:dim(your_data_subset1)[1]) {
if (i <dim(your_data_subset1)[1]){
#if it is the first annual weight of an individual, keep it
if (is.na(your_data_subset1$percdiff_turnpt[i])) {
your_data_subset2[i,] <- your_data_subset1[i,]
}
#for inflection points between the first and last annual weights for each individual
else {
#keep if it is truly an inflection point (aka, local maximum)
if (your_data_subset1$individual_ID[i] == your_data_subset1$individual_ID[i+1]) {
if (your_data_subset1$percdiff_turnpt[i] > 0 & your_data_subset1$percdiff_turnpt[i+1] < 0 ) {
your_data_subset2[i,] <- your_data_subset1[i,]
}
#keep if it is truly an inflection point (aka, local minimum)
else if (your_data_subset1$percdiff_turnpt[i] < 0 & your_data_subset1$percdiff_turnpt[i+1] > 0 ) {
your_data_subset2[i,] <- your_data_subset1[i,]
}
#otherwise it is a plateau point and set it to NA
#caveat: this may also remove a true inflection point depending on if the weight goes down for a local maximum plateau, or if the weight goes up for a local minimum plateau
#This is not an issue if the weight keeps going up after a local maximum plateau, or if the weight keeps going down for a local minimum plateau (which is the case for all plateau points in our dataset)
else {your_data_subset2[i,] <- your_data_subset2[i,]}
}
#if it is the last annual weight of an individual, keep it
else if (your_data_subset1$individual_ID[i]!= your_data_subset1$individual_ID[i+1]) {
your_data_subset2[i,] <- your_data_subset1[i,]
}
}
}
else if (i==dim(your_data_subset1)[1]) {
#if it is the last annual weight of the last individual, keep it
your_data_subset2[i,] <- your_data_subset1[i,]
}
}
your_data_subset2 <- your_data_subset2[which(!is.na(your_data_subset2$annualKG)),]
#using the code above, only the first, inflection points, and last annual weights per individual were kept, excluding those plateau points
#the rows with NA were removed
#Fifth (optional), recalculate the weight percentage difference between the corrected inflection points (annual weight at inflection point 2 - annual weight at inflection point 1)
#create the % wt change between each inflection point
test<-your_data_subset2
for (i in 1:dim(your_data_subset2)[1]) {
if (i>1){
#if it is the same individual, compute the weight different (in %) between the two inflection points
if (test[i,1]==test[i+1,1]){
test$percdiff_turnpt_v2[i+1] <- test$annualKG[i+1]/test$annualKG[i]*100-100
}
#if it is not the same individual (i.e., change from the last annual weight for jth individual to the first annual weight for the (j+1)th individual), leave the first row as NA for (j+1)th individual, as it has no preceding weight from the same individual to calculate the weight difference
else if (test[i,1]!=test[i+1,1]) {
test$percdiff_turnpt_v2[i+1] <- NA
}
}
#for the first annual weight of the first individual in the dataset, leave the first row as NA, and compute the weight difference for the first inflection point (compared to the first annual weight)
if (i==1){
test$percdiff_turnpt_v2[i] <- NA
test$percdiff_turnpt_v2[i+1] <- test$annualKG[i+1]/test$annualKG[i]*100-100
}
}
your_data_subset2 <- test
#Sixth, calling clinically meaningful weight change (5% or 10%) between the inflection points
#5% cutoff
for (i in 1:dim(your_data_subset2)[1]){
#if the weight difference is not NA
if (!is.na(your_data_subset2$percdiff_turnpt_v2[i])) {
#if the weight change is >= 5% (weight gain), then it is a clinically relevant weight change, marked as 1
if (your_data_subset2$percdiff_turnpt_v2[i]>=5) {
your_data_subset2$cutoff_v2[i] <- 1
}
#if the weight change is <= -5% (weight loss), then it is also a clinically relevant weight change, marked as 99
else if (your_data_subset2$percdiff_turnpt_v2[i]<=-5) {
your_data_subset2$cutoff_v2[i] <- 99
}
#otherwise it is not a clinically relevant weight change, marked as 0
else {your_data_subset2$cutoff_v2[i] <- 0}
}
#if the weight change value is NA, also marked as 0
else if (is.na(your_data_subset2$percdiff_turnpt_v2[i])) {
your_data_subset2$cutoff_v2[i] <- 0
}
}
#same as above, but with 10% as the cutoff
for (i in 1:dim(your_data_subset2)[1]){
#if the weight difference is not NA
if (!is.na(your_data_subset2$percdiff_turnpt_v2[i])) {
#if the weight change is >= 10% (weight gain), then it is a clinically relevant weight change, marked as 1
if (your_data_subset2$percdiff_turnpt_v2[i]>=10) {
your_data_subset2$big_cutoff_v2[i] <- 1
}
#if the weight change is <= -10% (weight loss), then it is also a clinically relevant weight change, marked as 99
else if (your_data_subset2$percdiff_turnpt_v2[i]<=-10) {
your_data_subset2$big_cutoff_v2[i] <- 99
}
#otherwise it is not a clinically relevant weight change, marked as 0
else {your_data_subset2$big_cutoff_v2[i] <- 0}
}
#if the weight change value is NA, also marked as 0
else if (is.na(your_data_subset2$percdiff_turnpt_v2[i])) {
your_data_subset2$big_cutoff_v2[i] <- 0
}
}
#Seventh, count the total number of weight changes (>5% or 10%) between inflection points
library(dplyr)
grp1<-group_by(your_data_subset2,ondividual_ID)
wtcycle_index <-as.data.frame(summarise(grp1,sum(cutoff_v2)))
colnames(wtcycle_index)[colnames(wtcycle_index)=="sum(cutoff_v2)"] <- "wtcycle_index_v2"
wtcycle_index_big <-as.data.frame(summarise(grp1,sum(big_cutoff_v2)))
colnames(wtcycle_index_big)[colnames(wtcycle_index_big)=="sum(big_cutoff_v2)"] <- "wtcycle_index_big_v2"
#1 loss, 1 gain = 99 + 1 =100;
#similarly
#101 - 1 loss, 2 gain; 102 - 1 loss, 3 gain; 103 - 1 loss, 4 gain
#199 - 2 loss, 1 gain; 200 - 2 loss, 2 gain; 201 - 2 loss, 3 gain; 202 - 2 loss, 4 gain
#298 - 3 loss, 1 gain, 299 - 3 loss, 2 gain, 300, 3 loss, 3 gain, 301, 3 loss, 4 gain; 302, 3 loss, 5 gain
#397 - 4 loss, 1 gain; 398 - 4 loss, 2 gain; 399 - 4 loss, 3 gain; 400 - 4 loss, 4 gain
#0 - no weight loss or gain
#1,2,3,4 - ranges from 1 weight gain to 4 weight gains
#99, 198, 297, 396 - ranges from 1 weight loss to 4 weight losses
#weight cycle (have both weight gain and loss >= 5%)
#could be either 100,101,102,103,199,200,201,202,298,299,300,301,302,397,398,399,or 400
#need to merge this weight cycle index with your dataset (e.g. your_data)
#each individual will have the same weight cycle index (either 5% or 10%) across all annual weight measures
#after merging the weight cycle index with your original annual weight dataset (e.g. your_data)
#Eighth, identify individuals through the LOCAL maximum/minimum method
#any individuals with these weight cycle index has at least 1 clinically relevant weight gain and 1 clinically relevant weight loss according to their weight cycle index
#thus these are the weight cyclers identified through the local minimum/maximum approach using inflection points
your_data$wtcycle_turnpt_v2 <- ifelse(your_data$wtcycle_index %in% c(100,101,102,103,199,200,201,202,298,299,300,301,302,397,398,399,400),1,0)
your_data$wtcycle_bigturnpt_v2 <- ifelse(your_data$wtcycle_index_big %in% c(100,101,102,103,199,200,201,202,298,299,300,301,302,397,398,399,400),1,0)
##### end of the (1) local minimum/maximum script #####
######### (2) global minimum/maximum ###########
#compute the weight cycle using the overall weight change ( weight change % to the global maximum and minimum annual weights per individual, and see if it is >= 5% or 10%)
#To achieve this, we first need to know the time sequence (if the maximum annual KG is before the minimum annual KG, or vice versa)
#get the year of the maximum annual weight
#some individuals may have the maximum annual weight at multiple time points, and if this is the case, we are interested in getting the year of the first maximum weight here
for (i in 1:dim(your_data)[1]) {
if (your_data$annualKG[i] == your_data$max_annualKG[i]) {
your_data$max_annualKG_year[i] <- your_data$YOMeasure[i]
}
else {your_data$max_annualKG_year[i] <- NA}
}
#to get the first year of maximum annual weights per individual, first to only keep the rows with maximum annual weights
tmp <- your_data[which(!is.na(your_data$max_annualKG_year)),]
max_annualKG_year <- unique(tmp[,c("individual_ID","max_annualKG_year")])
library(plyr)
ID_max_annualKG_year<-ddply(max_annualKG_year,.(individual_ID),nrow)
for (i in 1:dim(max_annualKG_year)[1]){
id <- max_annualKG_year[i,1]
year <- max_annualKG_year$max_annualKG_year[which(max_annualKG_year$individual_ID==id)]
keep <- which.min(year)
max_annualKG_year$firstmaxyear[i] <- year[keep]
}
#similarly, to get the year of the minimum annual weight (the first year if there are multiple)
for (i in 1:dim(your_data)[1]) {
if (your_data$annualKG[i] == your_data$min_annualKG[i]) {
your_data$min_annualKG_year[i] <- your_data$YOMeasure[i]
}
else {your_data$min_annualKG_year[i] <- NA}
}
#to get the first year of the minimum annual weights per individual, first to only keep the rows with the minimum annual weights
tmp <- your_data[which(!is.na(your_data$min_annualKG_year)),]
min_annualKG_year <- unique(tmp[,c("individual_ID","min_annualKG_year")])
library(plyr)
ID_min_annualKG_year<-ddply(min_annualKG_year,.(individual_ID),nrow)
for (i in 1:dim(min_annualKG_year)[1]){
id <- min_annualKG_year[i,1]
year <- min_annualKG_year$min_annualKG_year[which(min_annualKG_year$individual_ID==id)]
keep <- which.min(year)
min_annualKG_year$firstminyear[i] <- year[keep]
}
#now we get the first year of both maximum annual weight and minimum annual weight, so we can compare which occurred first for each individual
#Next, we also need to compute the second minimum annual weight before/after maximum annual weight and/or second maximum annual weight before/after minimum annual weight
tmp <- your_data
for (i in 1:dim(tmp[1])){
id <- tmp$individual_ID[i]
dat <- tmp[which(tmp$individual_ID==id),]
#when the maximum annual weight occurred before the minimum annual weight
if (tmp$firstmaxyear[i] < tmp$firstminyear[i]) {
beforemax <- dat[which(dat$YOMeasure < dat$firstmaxyear),]
if (dim(beforemax)[1] > 0) {
#find the second minimum annual weight before the maximum annual weight
keep1 <- which.min(beforemax$annualKG)
tmp$second_min_annualKG[i] <- beforemax$annualKG[keep1]
}
else {tmp$second_min_annualKG[i] <- NA}
aftermin <- dat[which(dat$YOMeasure > dat$firstminyear),]
if (dim(aftermin)[1] > 0) {
#find the second maximum annual weight after the minimum annual weight
keep2 <- which.max(aftermin$annualKG)
tmp$second_max_annualKG[i] <- aftermin$annualKG[keep2]
}
else {tmp$second_max_annualKG[i] <- NA}
# tmp$max_over_2min [i] <- tmp$max_annualKG[i]/tmp$second_min_annualKG[i]*100-100
# tmp$max2_over_min[i] <- tmp$second_max_annualKG[i]/tmp$min_annualKG[i]*100-100
}
#when the maximum annual weight occurred after the minimum annual weight
else if (tmp$firstmaxyear[i] > tmp$firstminyear[i]) {
beforemin <- dat[which(dat$YOMeasure < dat$firstminyear),]
if (dim(beforemin)[1] > 0) {
#find the second maximum annual weight before the minimum annual weight
keep1 <- which.max(beforemin$annualKG)
tmp$second_max_annualKG[i] <- beforemin$annualKG[keep1]
}
else {tmp$second_max_annualKG[i] <- NA}
aftermax <- dat[which(dat$YOMeasure > dat$firstmaxyear),]
if (dim(aftermax)[1] > 0) {
#find the second minimum annual weight after the maximum annual weight
keep2 <- which.min(aftermax$annualKG)
tmp$second_min_annualKG[i] <- aftermax$annualKG[keep2]
}
else {tmp$second_min_annualKG[i] <- NA}
# tmp$min_over_2max [i] <- tmp$min_annualKG[i]/tmp$second_max_annualKG[i]*100-100
# tmp$min2_over_max[i] <- tmp$second_min_annualKG[i]/tmp$max_annualKG[i]*100-100
}
}
your_data <- tmp
#What if the second minimum or maximum annual weights occurred between the years of maximum and minimum annual weights
#aka, the first and last annual weights happened to be the maximum and minimum annual weights or vice versa
#compute the 2nd minimum and maximum annual weights in between minimum and maximum annual weights
tmp <- your_data
for (i in 1:dim(tmp)[1]){
id <- tmp$individual_ID[i]
dat <- tmp[which(tmp$individual_ID==id),]
# when the maximum annual weight occurred before the minimum annual weight
if (tmp$firstmaxyear[i] < tmp$firstminyear[i]) {
between <- dat[which(dat$YOMeasure > dat$firstmaxyear & dat$YOMeasure < dat$firstminyear),]
if (dim(between)[1]> 0) {
#get the second maximum and minimum annual weights and their corresponding years in between the maximum and minimum annual weights
#which.max or which.min get the row index of the first maximum or minimum values (even if there are multiple maximum or minimum values present)
keep_max <- which.max(between$annualKG)
keep_min <- which.min(between$annualKG)
tmp$between_second_max_annualKG[i] <- between$annualKG[keep_max]
tmp$between_second_min_annualKG[i] <- between$annualKG[keep_min]
tmp$between_second_max_annualKG_year[i] <- between$YOMeasure[keep_max]
tmp$between_second_min_annualKG_year[i] <- between$YOMeasure[keep_min]
}
else if (dim(between)[1] == 0) {
tmp$between_second_max_annualKG[i] <- NA
tmp$between_second_min_annualKG[i] <- NA
tmp$between_second_max_annualKG_year[i] <- NA
tmp$between_second_min_annualKG_year[i] <- NA
}
}
# when the maximum annual weight occurred after the minimum annual weight
else if (tmp$firstmaxyear[i] > tmp$firstminyear[i]) {
between <- dat[which(dat$YOMeasure < dat$firstmaxyear & dat$YOMeasure > dat$firstminyear),]
if (dim(between)[1]> 0) {
#get the second maximum and minimum annual weights and their corresponding years in between the maximum and minimum annual weights
keep_max <- which.max(between$annualKG)
keep_min <- which.min(between$annualKG)
tmp$between_second_max_annualKG[i] <- between$annualKG[keep_max]
tmp$between_second_min_annualKG[i] <- between$annualKG[keep_min]
tmp$between_second_max_annualKG_year[i] <- between$YOMeasure[keep_max]
tmp$between_second_min_annualKG_year[i] <- between$YOMeasure[keep_min]
}
else if (dim(between)[1] == 0) {
tmp$between_second_max_annualKG[i] <- NA
tmp$between_second_min_annualKG[i] <- NA
tmp$between_second_max_annualKG_year[i] <- NA
tmp$between_second_min_annualKG_year[i] <- NA
}
}
}
your_data <- tmp
#so now we get the years of maximum, second maximum, minimum and second minimum annual weights, so we can compare the time sequence of these annual weights for each individual
#Below are the several most common scenarios
#Situation 1: when max annual KG is before min annual KG
###Situation 1.1: 2nd min -> max -> min
###Situation 1.2: max -> min -> 2nd max
###Situation 1.3: 2nd min -> max -> min -> 2nd max
###Situation 1.4: max -> 2nd min -> 2nd max -> min
#Situation 2: when min annual KG is before max annual KG
###Situation 2.1: 2nd max -> min -> max
###Situation 2.2: min -> max -> 2nd min
###Situation 2.3: 2nd max -> min -> max -> 2nd min
###Situation 2.4: min -> 2nd max -> 2nd min -> max
#compute the weight change (%) from second maximum or minimum values to minimum or maximum annual weights, or vice versa
# (i) when the second maximum and minimum annual weights are outside the time interval of maximum and minimum annual weights
#when the second maximum annual weight occurred before the minimum annual weight
your_data$min_over_max2<- your_data$min_annualKG/your_data$second_max_annualKG*100-100
#when the second maximum annual weight occurred after the minimum annual weight
your_data$max2_over_min <- your_data$second_max_annualKG/your_data$min_annualKG*100-100
#when the second minimum annual weight occurred after the maximum annual weight
your_data$min2_over_max <- your_data$second_min_annualKG/your_data$max_annualKG*100-100
#when the second minimum annual weight occurred before the maximum annual weight
your_data$max_over_min2 <- your_data$max_annualKG/your_data$second_min_annualKG*100-100
# (ii) when the second maximum and minimum annual weights are within the time interval of maximum and minimum annual weights
#when the second maximum annual weight occurred after the second minimum annual weight
your_data$max2_over_min2 <- your_data$between_second_max_annualKG/your_data$between_second_min_annualKG*100-100
#when the second minimum annual weight occurred after the second maximum annual weight
your_data$min2_over_max2 <- your_data$between_second_min_annualKG/your_data$between_second_max_annualKG*100-100
#when the second maximum annual weight occurred before the minimum annual weight
your_data$between_min_over_max2<- your_data$min_annualKG/your_data$between_second_max_annualKG*100-100
#when the second maximum annual weight occurred after the minimum annual weight
your_data$between_max2_over_min <- your_data$between_second_max_annualKG/your_data$min_annualKG*100-100
#when the second minimum annual weight occurred after the maximum annual weight
your_data$between_min2_over_max <- your_data$between_second_min_annualKG/your_data$max_annualKG*100-100
#when the second minimum annual weight occurred before the maximum annual weight
your_data$between_max_over_min2 <- your_data$max_annualKG/your_data$between_second_min_annualKG*100-100
# (i) model the weight cycle when the second maximum and minimum annual weights are outside the time interval of maximum and minimum annual weights
#compute the wtcycle using max, min and second max/min before or after the max/min values
your_data$perc_maxmin_annualKG <- your_data$max_annualKG/your_data$min_annualKG*100-100
your_data$perc_minmax_annualKG <- your_data$min_annualKG/your_data$max_annualKG*100-100
tmp <- your_data
#5% cutoff
for (i in 1:dim(tmp)[1]) {
#when the maximum annual weight occurred before the minimum annual weight
if (tmp$firstmaxyear[i] < tmp$firstminyear[i]) {
#maximum annual weight is the first annual weight and minimum annual weight is the last annual weight
if (is.na(tmp$second_min_annualKG[i]) & is.na(tmp$second_max_annualKG[i])) {
tmp$wtcycle_maxmin_5[i] <- 0
}
#Situation 1.1: 2nd min -> max -> min
else if (!is.na(tmp$second_min_annualKG[i]) & is.na(tmp$second_max_annualKG[i])) {
if (tmp$max_over_min2[i] >= 5 & tmp$perc_minmax_annualKG[i] <= -5) {
tmp$wtcycle_maxmin_5[i] <- 1
}
else { tmp$wtcycle_maxmin_5[i] <- 0}
}
#Situation 1.2: max -> min -> 2nd max
else if (is.na(tmp$second_min_annualKG[i]) & !is.na(tmp$second_max_annualKG[i])) {
if (tmp$max2_over_min[i] >= 5 & tmp$perc_minmax_annualKG[i] <= -5) {
tmp$wtcycle_maxmin_5[i] <- 1
}
else {tmp$wtcycle_maxmin_5[i] <- 0}
}
#Situation 1.3: 2nd min -> max -> min -> 2nd max
else if (!is.na(tmp$second_min_annualKG[i]) & !is.na(tmp$second_max_annualKG[i])) {
#if the change from maximum to minimum annual weight was greater than 5%
#AND any change on either of the two ends (2nd min -> max , or min -> 2nd max) was greater than 5% in Situation 1.3
if (tmp$perc_minmax_annualKG[i] <= -5 & (tmp$max2_over_min[i] >= 5 | tmp$max_over_min2[i] >= 5 )) {
tmp$wtcycle_maxmin_5[i] <- 1
}
else {tmp$wtcycle_maxmin_5[i] <- 0}
}
}
#when the maximum annual weight occurred after the minimum annual weight
else if (tmp$firstmaxyear[i] > tmp$firstminyear[i]) {
#maximum annual weight is the last annual weight and minimum annual weight is the first annual weight
if (is.na(tmp$second_min_annualKG[i]) & is.na(tmp$second_max_annualKG[i])) {
tmp$wtcycle_maxmin_5[i] <- 0
}
#Situation 2.1: 2nd max -> min -> max
else if (is.na(tmp$second_min_annualKG[i]) & !is.na(tmp$second_max_annualKG[i])) {
if (tmp$min_over_max2[i] <= -5 & tmp$perc_maxmin_annualKG[i] >= 5) {
tmp$wtcycle_maxmin_5[i] <- 1
}
else {tmp$wtcycle_maxmin_5[i] <- 0}
}
#Situation 2.2: min -> max -> 2nd min
else if (!is.na(tmp$second_min_annualKG[i]) & is.na(tmp$second_max_annualKG[i])) {
if (tmp$min2_over_max[i] <= -5 & tmp$perc_maxmin_annualKG[i] >= 5) {
tmp$wtcycle_maxmin_5[i] <- 1
}
else { tmp$wtcycle_maxmin_5[i] <- 0}
}
#Situation 2.3: 2nd max -> min -> max -> 2nd min
else if (!is.na(tmp$second_min_annualKG[i]) & !is.na(tmp$second_max_annualKG[i])) {
#if the change from minimum to maximum annual weight was greater than 5%
#AND any change on either of the two ends (2nd max -> min , or max -> 2nd min) was greater than 5% in Situation 2.3
if (tmp$perc_maxmin_annualKG[i] >= 5 & (tmp$min2_over_max[i] <= -5 | tmp$min_over_max2[i] <= -5 )) {
tmp$wtcycle_maxmin_5[i] <- 1
}
else {tmp$wtcycle_maxmin_5[i] <- 0}
}
}
}
#10% cutoff
for (i in 1:dim(tmp)[1]) {
#when the maximum annual weight occurred before the minimum annual weight
if (tmp$firstmaxyear[i] < tmp$firstminyear[i]) {
#maximum annual weight is the first annual weight and minimum annual weight is the last annual weight
if (is.na(tmp$second_min_annualKG[i]) & is.na(tmp$second_max_annualKG[i])) {
tmp$wtcycle_maxmin_10[i] <- 0
}
#Situation 1.1: 2nd min -> max -> min
else if (!is.na(tmp$second_min_annualKG[i]) & is.na(tmp$second_max_annualKG[i])) {
if (tmp$max_over_min2[i] >= 10 & tmp$perc_minmax_annualKG[i] <= -10) {
tmp$wtcycle_maxmin_10[i] <- 1
}
else { tmp$wtcycle_maxmin_10[i] <- 0}
}
#Situation 1.2: max -> min -> 2nd max
else if (is.na(tmp$second_min_annualKG[i]) & !is.na(tmp$second_max_annualKG[i])) {
if (tmp$max2_over_min[i] >= 10 & tmp$perc_minmax_annualKG[i] <= -10) {
tmp$wtcycle_maxmin_10[i] <- 1
}
else {tmp$wtcycle_maxmin_10[i] <- 0}
}
#Situation 1.3: 2nd min -> max -> min -> 2nd max
else if (!is.na(tmp$second_min_annualKG[i]) & !is.na(tmp$second_max_annualKG[i])) {
#if the change from maximum to minimum annual weight was greater than 10%
#AND any change on either of the two ends (2nd min -> max , or min -> 2nd max) was greater than 10% in Situation 1.3
if (tmp$perc_minmax_annualKG[i] <= -10 & (tmp$max2_over_min[i] >= 10 | tmp$max_over_min2[i] >= 10 )) {
tmp$wtcycle_maxmin_10[i] <- 1
}
else {tmp$wtcycle_maxmin_10[i] <- 0}
}
}
#when the maximum annual weight occurred after the minimum annual weight
else if (tmp$firstmaxyear[i] > tmp$firstminyear[i]) {
#maximum annual weight is the last annual weight and minimum annual weight is the first annual weight
if (is.na(tmp$second_min_annualKG[i]) & is.na(tmp$second_max_annualKG[i])) {
tmp$wtcycle_maxmin_10[i] <- 0
}
#Situation 2.1: 2nd max -> min -> max
else if (is.na(tmp$second_min_annualKG[i]) & !is.na(tmp$second_max_annualKG[i])) {
if (tmp$min_over_max2[i] <= -10 & tmp$perc_maxmin_annualKG[i] >= 10) {
tmp$wtcycle_maxmin_10[i] <- 1
}
else {tmp$wtcycle_maxmin_10[i] <- 0}
}
#Situation 2.2: min -> max -> 2nd min
else if (!is.na(tmp$second_min_annualKG[i]) & is.na(tmp$second_max_annualKG[i])) {
if (tmp$min2_over_max[i] <= -10 & tmp$perc_maxmin_annualKG[i] >= 10) {
tmp$wtcycle_maxmin_10[i] <- 1
}
else { tmp$wtcycle_maxmin_10[i] <- 0}
}
#Situation 2.3: 2nd max -> min -> max -> 2nd min
else if (!is.na(tmp$second_min_annualKG[i]) & !is.na(tmp$second_max_annualKG[i])) {
#if the change from minimum to maximum annual weight was greater than 10%
#AND any change on either of the two ends (2nd max -> min , or max -> 2nd min) was greater than 10% in Situation 2.3
if (tmp$perc_maxmin_annualKG[i] >= 10 & (tmp$min2_over_max[i] <= -10 | tmp$min_over_max2[i] <= -10 )) {
tmp$wtcycle_maxmin_10[i] <- 1
}
else {tmp$wtcycle_maxmin_10[i] <- 0}
}
}
}
your_data <- tmp
# (ii) Model the weight cycle using the global maximum/minimum approach when the second maximum and minimum annual weights are within the time interval of maximum and minimum annual weights
tmp <- your_data
#5% cutoff
for (i in 1:dim(tmp)[1]){
if (!is.na(tmp$between_second_min_annualKG[i])) {
#when the maximum annual weight occurred before the minimum annual weight
if (tmp$firstmaxyear[i] < tmp$firstminyear[i]) {
#Situation 1.4: max -> 2nd min -> 2nd max -> min
if (tmp$between_second_min_annualKG_year[i] < tmp$between_second_max_annualKG_year[i]) {
#if the weight gain and loss were both greater than 5% for this part (max -> 2nd min -> 2nd max) in Situation 1.4
if ( tmp$between_min2_over_max[i] <= -5 & tmp$max2_over_min2[i] >= 5) {
tmp$wtcycle_between_5[i] <- 1
}
#if the weight gain and loss were both greater than 5% for this part (2nd min -> 2nd max -> min ) in Situation 1.4
else if ( tmp$max2_over_min2[i] >= 5 & tmp$between_min_over_max2[i] <= -5) {
tmp$wtcycle_between_5[i] <- 1
}
else { tmp$wtcycle_between_5[i] <- 0}
}
else {tmp$wtcycle_between_5[i] <- 0}
}
#when the maximum annual weight occurred after the minimum annual weight
else if (tmp$firstmaxyear[i] > tmp$firstminyear[i]) {
#Situation 2.4: min -> 2nd max -> 2nd min -> max
if (tmp$between_second_max_annualKG_year[i] < tmp$between_second_min_annualKG_year[i]) {
#if the weight gain and loss were both greater than 5% for this part ( min -> 2nd max -> 2nd min) in Situation 2.4
if ( tmp$between_max2_over_min[i] >=5 & tmp$min2_over_max2[i] <= -5) {
tmp$wtcycle_between_5[i] <- 1
}
#if the weight gain and loss were both greater than 5% for this part ( 2nd max -> 2nd min -> max ) in Situation 2.4
else if ( tmp$min2_over_max2[i] <= -5 & tmp$between_max_over_min2[i] >= 5) {
tmp$wtcycle_between_5[i] <- 1
}
else { tmp$wtcycle_between_5[i] <- 0}
}
else
{ tmp$wtcycle_between_5[i] <- 0}
}
}
else if (is.na(tmp$between_second_min_annualKG[i])) {
tmp$wtcycle_between_5[i] <- 0
}
}
#10% cutoff
for (i in 1:dim(tmp)[1]){
if (!is.na(tmp$between_second_min_annualKG[i])) {
#when the maximum annual weight occurred before the minimum annual weight
if (tmp$firstmaxyear[i] < tmp$firstminyear[i]) {
#Situation 1.4: max -> 2nd min -> 2nd max -> min
if (tmp$between_second_min_annualKG_year[i] < tmp$between_second_max_annualKG_year[i]) {
#if the weight gain and loss were both greater than 10% for this part (max -> 2nd min -> 2nd max) in Situation 1.4
if ( tmp$between_min2_over_max[i] <= -10 & tmp$max2_over_min2[i] >= 10) {
tmp$wtcycle_between_10[i] <- 1
}
#if the weight gain and loss were both greater than 10% for this part (2nd min -> 2nd max -> min ) in Situation 1.4
else if ( tmp$max2_over_min2[i] >= 10 & tmp$between_min_over_max2[i] <= -10) {
tmp$wtcycle_between_10[i] <- 1
}
else { tmp$wtcycle_between_10[i] <- 0}
}
else {tmp$wtcycle_between_10[i] <- 0}
}
#when the maximum annual weight occurred after the minimum annual weight
else if (tmp$firstmaxyear[i] > tmp$firstminyear[i]) {
#Situation 2.4: min -> 2nd max -> 2nd min -> max
if (tmp$between_second_max_annualKG_year[i] < tmp$between_second_min_annualKG_year[i]) {
#if the weight gain and loss were both greater than 10% for this part ( min -> 2nd max -> 2nd min) in Situation 2.4
if ( tmp$between_max2_over_min[i] >=10 & tmp$min2_over_max2[i] <= -10) {
tmp$wtcycle_between_10[i] <- 1
}
#if the weight gain and loss were both greater than 10% for this part ( 2nd max -> 2nd min -> max ) in Situation 2.4
else if ( tmp$min2_over_max2[i] <= -10 & tmp$between_max_over_min2[i] >= 10) {
tmp$wtcycle_between_10[i] <- 1
}
else { tmp$wtcycle_between_10[i] <- 0}
}
else
{ tmp$wtcycle_between_10[i] <- 0}
}
}
else if (is.na(tmp$between_second_min_annualKG[i])) {
tmp$wtcycle_between_10[i] <- 0
}
}
your_data <- tmp
##### end of the (2) global minimum/maximum #######
### FINALLY, create the weight cycle trajectory binary variable #####
#if 5% is used as the cutoff for clinically meaningful weight change
for (i in 1:dim(your_data)[1]){
if (your_data$wtcycle_turnpt_v2[i]==1 | your_data$wtcycle_maxmin_5[i]==1|your_data$wtcycle_between_5[i]==1) {
your_data$weight_cycle_5[i] <- 1
}
else {your_data$weight_cycle_5[i] <- 0}
}
#if 10% is used as the cutoff for clinically meaningful weight change
for (i in 1:dim(your_data)[1]){
if (your_data$wtcycle_bigturnpt_v2[i]==1 | your_data$wtcycle_maxmin_10[i]==1|your_data$wtcycle_between_10[i]==1) {
your_data$weight_cycle_10[i] <- 1
}
else {your_data$weight_cycle_10[i] <- 0}
}
######### completion of weight cycle classification! ############
################# Check the summary statistics of each weight trajectory ##################
############################################################################################
table(your_data$stable_weight_5)
table(your_data$stable_weight_10)
table(your_data$weight_loss_5)
table(your_data$weight_loss_10)
table(your_data$weight_gain_5)
table(your_data$weight_gain_10)
table(your_data$weight_cycle_5)
table(your_data$weight_cycle_10)
############################################################################################