big.Rmd

We systematically explore the data in the MIMIC III database, focusing on labevents and chartevents where the values are numeric. To simplify and speed up this process, we read all the relevant data into R’s memory. This only works on machines with very large RAM provisioning. I believe it requires a system with 64GB of RAM, and have verified that it does not work on a system with “only” 32GB.

library(tidyverse)
library(RMySQL)
library(lubridate)
con <- dbConnect(MySQL(), dbname = "mimic3v14", username="mimic3", password=pwd, host=host)
in_clause <- function(items, col="itemid") {
  if (class(items)=="numeric")
    paste(col, " in (", paste(items, collapse=","), ")", sep="")
  else if (class(items)=="character")
    paste(col, "in (", paste0(sprintf("'%s'", items), collapse = ", "), ")")
}

Redundancy among different uses of itemid

We have previously checked to see whether there is any overlap between the same itemids appearing in different tables that use them. There are no overlaps in itemids among the following tables, so at least for this version of MIMIC, we do not need to check again.

• labevents
• chartevents
• datetimeevents
• inputevents_cv
• inputevents_mv
• outputevents
• procedureevents_mv
• microbiologyevents

Note that for microbiologyevents, itemids are used in three fields, spec_itemid, org_itemid, and ab_itemid. They also do not overlap each other.

Read needed data

Because all the data require a large memory, we can debug the code by reading only a subset. In order to sample uniformly from the available data, we take only every 5th sample from labevents and every 30th from chartevents.

First, all the labs

# labs <- dbReadTable(con, "labevents")
sql_labs <- "select subject_id, hadm_id, itemid, charttime, valuenum, valueuom, flag from labevents where valuenum is not null"
# sql_labs <- "select subject_id, hadm_id, itemid, charttime, valuenum, valueuom, flag from labevents where valuenum is not null and mod(row_id, 5)=0"
labs <- dbGetQuery(con, sql_labs)
labs$charttime <- ymd_hms(labs$charttime)
labs$valueuom <- factor(labs$valueuom)
labs$flag <- factor(labs$flag)
summary(labs)

   subject_id       hadm_id            itemid        charttime                      valuenum           valueuom             flag         
 Min.   :    2   Min.   :100001    Min.   :50801   Min.   :2096-02-26 01:00:00   Min.   :    -414   mEq/L  :5326273   abnormal: 9740677  
 1st Qu.:11362   1st Qu.:125117    1st Qu.:50882   1st Qu.:2126-07-26 03:54:00   1st Qu.:       4   mg/dL  :5038921   delta   :   64571  
 Median :22951   Median :149888    Median :50971   Median :2151-02-12 03:01:00   Median :      18   %      :3942299   NA's    :15127587  
 Mean   :32129   Mean   :149976    Mean   :51046   Mean   :2151-06-04 04:16:10   Mean   :      79   K/uL   :1533311                      
 3rd Qu.:49998   3rd Qu.:175035    3rd Qu.:51249   3rd Qu.:2176-07-19 04:44:00   3rd Qu.:      60   IU/L   :1025101                      
 Max.   :99999   Max.   :199999    Max.   :51555   Max.   :2210-08-24 05:53:00   Max.   :14272000   (Other):7059667                      
                 NA's   :4817521                                                                    NA's   :1007263                      

Then the chart events

# charts <- dbReadTable(con, "chartevents")
sql_charts <- "select subject_id, hadm_id, icustay_id, itemid, charttime, storetime, cgid, valuenum, valueuom, warning, error, resultstatus, stopped from chartevents where valuenum is not null"
# sql_charts <- "select subject_id, hadm_id, icustay_id, itemid, charttime, storetime, cgid, valuenum, valueuom, warning, error, resultstatus, stopped from chartevents where valuenum is not null and mod(row_id, 30)=0"
charts <- dbGetQuery(con, sql_charts)
charts$charttime <- ymd_hms(charts$charttime)
charts$storetime <- ymd_hms(charts$storetime)
charts$valueuom <- factor(charts$valueuom)
charts$warning <- factor(charts$warning)
charts$error <- factor(charts$error)
charts$resultstatus <- factor(charts$resultstatus)
charts$stopped <- factor(charts$stopped)
summary(charts)

   subject_id       hadm_id         icustay_id         itemid         charttime                     storetime                        cgid            valuenum       
 Min.   :    2   Min.   :100001   Min.   :200001   Min.   :     2   Min.   :2100-06-07 20:00:00   Min.   :2100-06-07 20:00:00   Min.   :14010     Min.   :  -10069  
 1st Qu.:11043   1st Qu.:125400   1st Qu.:225957   1st Qu.:   646   1st Qu.:2127-10-14 20:00:00   1st Qu.:2126-09-27 19:44:00   1st Qu.:15449     1st Qu.:       7  
 Median :22378   Median :149029   Median :250823   Median :  5814   Median :2152-12-04 15:00:00   Median :2151-06-24 23:07:00   Median :17474     Median :      40  
 Mean   :30873   Mean   :149782   Mean   :250393   Mean   : 75025   Mean   :2152-03-26 20:51:45   Mean   :2151-07-07 05:16:42   Mean   :17624     Mean   :      72  
 3rd Qu.:46092   3rd Qu.:174660   3rd Qu.:275019   3rd Qu.:220181   3rd Qu.:2177-07-29 19:00:00   3rd Qu.:2176-11-14 15:51:00   3rd Qu.:19648     3rd Qu.:      96  
 Max.   :99999   Max.   :199999   Max.   :299999   Max.   :228444   Max.   :2209-08-07 16:52:00   Max.   :2209-08-07 16:52:00   Max.   :21570     Max.   :10000000  
                                  NA's   :147506                                                  NA's   :6312672               NA's   :6312672                     
    valueuom        warning          error          resultstatus           stopped        
 mmHg   :30642461   0   :45893601   0   :47922529   Final :  8739290   NotStopd:98323835  
 BPM    :15386992   1   : 2091540   1   :   62612   Manual:   665998   NA's    :48180333  
 %      :13896202   NA's:98519027   NA's:98519027   NA's  :137098880                      
 bpm    : 8317335                                                                         
 kg     : 6633378                                                                         
 (Other):41584934                                                                         
 NA's   :30042866                                                                         

Then we see how often each of the lab and chart items occur in our data

d_labitems <- dbReadTable(con, "d_labitems")
names(d_labitems) <- tolower(names(d_labitems))
d_labitems$row_id <- NULL
d_labitems$category <- tools::toTitleCase(tolower(d_labitems$category)) # fixes a few capitalization errors in data
# d_labitems$label <- factor(d_labitems$label)
d_labitems$fluid <- factor(d_labitems$fluid)
d_labitems$category <- factor(d_labitems$category)
d_labitems <- labs %>% 
  group_by(itemid) %>% 
  summarize(n=n(), n_subjects=n_distinct(subject_id), n_adm=n_distinct(hadm_id)) %>%
  arrange(desc(n)) %>% 
  ungroup() %>%
  inner_join(d_labitems, by="itemid")
d_labitems$desc <- 
  sprintf("%d (%d) %s in %s%s",
          d_labitems$itemid, d_labitems$n, d_labitems$label, d_labitems$fluid,
          ifelse(d_labitems$category=="Blood Gas", " {BG}",
                 ifelse(d_labitems$category=="Chemistry", " {Chem}",
                        " {Hem}")))
d_labitems$linksto <- factor("labevents")
summary(d_labitems)

     itemid            n              n_subjects           n_adm            label                                 fluid           category    loinc_code       
 Min.   :50801   Min.   :     1.0   Min.   :    1.00   Min.   :    1.0   Length:494         Blood                    :250   Blood Gas : 29   Length:494        
 1st Qu.:50961   1st Qu.:    31.5   1st Qu.:   24.25   1st Qu.:   17.0   Class :character   Urine                    : 64   Chemistry :226   Class :character  
 Median :51106   Median :   426.0   Median :  285.50   Median :  221.5   Mode  :character   Other Body Fluid         : 49   Hematology:239   Mode  :character  
 Mean   :51134   Mean   : 50471.3   Mean   : 5258.53   Mean   : 5925.8                      Ascites                  : 36                                      
 3rd Qu.:51299   3rd Qu.:  4697.2   3rd Qu.: 2228.25   3rd Qu.: 1738.0                      Pleural                  : 34                                      
 Max.   :51555   Max.   :881653.0   Max.   :45383.00   Max.   :57103.0                      Cerebrospinal Fluid (CSF): 27                                      
                                                                                            (Other)                  : 34                                      
     desc                linksto   
 Length:494         labevents:494  
 Class :character                  
 Mode  :character                  
                                   
                                   
                                   
                                   

Correlate the above counts with information on each of the items from the d_items and d_labitems tables

d_items <- dbReadTable(con, "d_items")
names(d_items) <- tolower(names(d_items))
d_items$row_id <- NULL
d_items$conceptid <- NULL # this column is all nulls
# d_items$label <- factor(d_items$label)
d_items$abbreviation <- factor(d_items$abbreviation)
d_items$dbsource <- factor(d_items$dbsource)
d_items$linksto <- factor(d_items$linksto)
d_items$category <- factor(d_items$category)
d_items$unitname <- factor(d_items$unitname)
d_items$param_type <- factor(d_items$param_type)
d_items <- charts %>% 
  group_by(itemid) %>% 
  summarize(n=n(), n_subjects=n_distinct(subject_id), n_adm=n_distinct(hadm_id)) %>% 
  arrange(desc(n)) %>%
  ungroup() %>%
  inner_join(d_items, by="itemid")
d_items$desc <- 
  sprintf("%d-%s (%d) %s %s",
          d_items$itemid, 
          ifelse(d_items$dbsource=="hospital", "hosp",
             ifelse(d_items$dbsource=="carevue", "cv",
                    ifelse(d_items$dbsource=="metavision", "mv", "???"))), 
          d_items$n, d_items$label,
          ifelse(is.na(d_items$category), "", paste0("{", d_items$category, "}", sep="")))
summary(d_items)

     itemid             n             n_subjects          n_adm            label                                        abbreviation        dbsource   
 Min.   :     2   Min.   :      1   Min.   :    1.0   Min.   :    1.0   Length:2884        14 Gauge Dressing Occlusive        :   1   carevue   :2111  
 1st Qu.:  2024   1st Qu.:      8   1st Qu.:    1.0   1st Qu.:    1.0   Class :character   14 Gauge placed in outside facility:   1   hospital  :   0  
 Median :  5578   Median :     64   Median :    5.0   Median :    5.0   Mode  :character   14 Gauge placed in the field       :   1   metavision: 773  
 Mean   : 63171   Mean   :  50799   Mean   : 1770.6   Mean   : 2039.7                      16 Gauge Dressing Occlusive        :   1                    
 3rd Qu.:220671   3rd Qu.:   4369   3rd Qu.:  510.8   3rd Qu.:  537.5                      16 Gauge placed in outside facility:   1                    
 Max.   :228444   Max.   :5179363   Max.   :29898.0   Max.   :34901.0                      (Other)                            : 768                    
                                                                                           NA's                               :2111                    
               linksto                        category       unitname               param_type       desc          
 chartevents       :2884   Labs                   :  98   None   : 196   Numeric         : 450   Length:2884       
 datetimeevents    :   0   Access Lines - Invasive:  74   mmHg   :  72   Checkbox        : 190   Class :character  
 inputevents_cv    :   0   Respiratory            :  67   cm     :  35   Text            : 100   Mode  :character  
 inputevents_mv    :   0   Chemistry              :  56   %      :  30   Numeric with tag:  33                     
 microbiologyevents:   0   Scores - APACHE IV (2) :  53   L/min  :  21   Date time       :   0                     
 outputevents      :   0   (Other)                : 621   (Other): 129   (Other)         :   0                     
 procedureevents_mv:   0   NA's                   :1915   NA's   :2401   NA's            :2111                     

Combine items and labitems

d_all <- rbind(
  d_labitems[,c("itemid", "n", "n_subjects", "n_adm", "label", "category", "linksto", "desc")],
  d_items[,c("itemid", "n", "n_subjects", "n_adm", "label", "category", "linksto", "desc")]
) %>% arrange(desc(n))
summary(d_all)

     itemid             n             n_subjects          n_adm            label                              category                  linksto         desc          
 Min.   :     2   Min.   :      1   Min.   :    1.0   Min.   :    1.0   Length:3378        Chemistry              : 282   chartevents       :2884   Length:3378       
 1st Qu.:  2358   1st Qu.:      9   1st Qu.:    1.0   1st Qu.:    1.0   Class :character   Hematology             : 254   labevents         : 494   Class :character  
 Median :  6539   Median :     81   Median :   10.0   Median :    9.5   Mode  :character   Labs                   :  98   datetimeevents    :   0   Mode  :character  
 Mean   : 61410   Mean   :  50751   Mean   : 2280.7   Mean   : 2608.0                      Access Lines - Invasive:  74   inputevents_cv    :   0                     
 3rd Qu.: 51428   3rd Qu.:   4398   3rd Qu.:  796.8   3rd Qu.:  771.8                      Respiratory            :  67   inputevents_mv    :   0                     
 Max.   :228444   Max.   :5179363   Max.   :45383.0   Max.   :57103.0                      (Other)                : 688   microbiologyevents:   0                     
                                                                                           NA's                   :1915   (Other)           :   0                     

all <- rbind(labs %>% select(subject_id, hadm_id, itemid, charttime, valuenum, valueuom),
             charts %>% select(subject_id, hadm_id, itemid, charttime, valuenum, valueuom))

Find all chart and lab events with the same labels.

We assume that data elements with identical labels may be related to each other, and will investigate this by comparing distributions of values. In some instances, the same label may identify measurements from different types of samples (e.g., blood vs. urine) or by different techniques, so we do not expact all the distributions to align.

d_bylabels <- d_all %>%
  group_by(label) %>%
  summarize(ntot=sum(n),
            ambig=n(),
            ns=paste(n, collapse=", "),
            itemids=paste0(itemid, collapse=", "),
            categories=paste0(category, collapse=", "),
            # dbsources=paste0(dbsource, collapse=", "),
            linksto=paste0(linksto, collapse=", ")
            # units=paste0(unitname, collapse=", "),
            # param_types=paste0(param_type, collapse=", ")
            ) %>%
  ungroup() %>%
  arrange(label)
d_bylabels_ambig <- d_bylabels %>% filter(ambig>1) %>% arrange(label)
d_bylabels_ambig

There are 3190 distinct labels for events, of which 119 are associated with multiple itemids. We consider the distribution of numerical values of each of these sets, though we limit examination to labels with a total count of at least a threshold number. We also eliminate outliers by choosing some fraction of the extremes of each distribution to omit. (We begin by eliminating only 1%, i.e., 0.5% at each extreme. This seems to get rid of most artifact values such a negative lab measurements or ridiculously high ones.)

For each distinct label associated with multiple itemids, we plot distributions of their values separately and overplotted on each other, plus violin plots and cumulative distributions to allow different ways to visualize the relationships between the distributions of each itemid’s data.

total_count_threshold = 1000
outliers_fraction = 0.01
check_for_duplicates <- TRUE
d_chart_ambig_common <- d_bylabels_ambig %>% filter(ntot>=total_count_threshold) %>% arrange(desc(ntot))
print(d_chart_ambig_common)

bin_width <- function(dat) {
  log_range <- log10(max(dat) - min(dat))
  binw <- ifelse(log_range < 0.0, .01,
                 ifelse(log_range < 1.0, 0.1,
                        ifelse(log_range < 2.2, 1,
                               ifelse(log_range < 3.2, 10, 100))))
}
rid_outliers <- function(d, outliers_fraction = 0.01) {
  if (outliers_fraction != 0.0) {
    limits <- quantile(d$valuenum, probs=c(outliers_fraction/2, 1-outliers_fraction/2), na.rm=TRUE)
    d %>% filter(valuenum <= limits[2] & valuenum >= limits[1])
  } else {
    d
  }
}
dups <- NULL
for (l in 1:nrow(d_chart_ambig_common)) {
  # print(l)
  lbldata <- d_chart_ambig_common[l,]
  items <- (d_all %>% filter(label==lbldata$label))$itemid
  # print(items)
  
  dat <- all %>% filter(itemid %in% items) %>%
    left_join(d_all, by="itemid")
  
  if (outliers_fraction != 0.0) {
    limits <- quantile(dat$valuenum, probs=c(outliers_fraction/2, 1-outliers_fraction/2), na.rm=TRUE)
    dat <- dat %>% filter(valuenum <= limits[2] & valuenum >= limits[1])
  }
  binw <- bin_width(dat$valuenum)
  
  print(dat %>%
          ggplot(aes(valuenum, fill=desc)) + 
          geom_histogram(binwidth = binw) +
          facet_wrap(~ desc) +
          theme(legend.position="none") +
          ggtitle(lbldata$label))
  print(dat %>%
          ggplot(aes(valuenum, fill=desc)) +
          geom_histogram(binwidth = binw, alpha=0.5) +
          ggtitle(lbldata$label) +
          theme(legend.position="bottom", legend.direction="vertical"))
  print(dat %>%
          ggplot(aes(x=desc, y=valuenum, color=desc)) +
          geom_violin() +
          coord_flip() +
          theme(legend.position="none") +
          ggtitle(lbldata$label))
  print(dat %>%
          ggplot(aes(valuenum, color=desc)) +
          stat_ecdf(aes(linetype = desc)) +
          theme(legend.position="bottom", legend.direction="vertical") +
          ggtitle(sprintf("CDF of %s", lbldata$label)))
  # Collect duplication data while producing plots
  if (check_for_duplicates) {
    mults <- dat %>%
      group_by(subject_id, charttime) %>%
      summarize(
        itemids = paste(unique(itemid), collapse=", "),
        nitemids = n_distinct(itemid),
        nvals = n(),
        ndistinct = n_distinct(valuenum),
        vals = paste(valuenum, collapse=", "),
        val_mean = mean(valuenum),
        val_low = min(valuenum),
        val_high = max(valuenum)
      ) %>%
      ungroup() %>%
      filter(nvals > 1)
    if (nrow(mults) > 0) {
      dups <- rbind(dups, mults)
    }
  }
}

Duplicate values

We have observed that sometimes among the itemids associated with a particular label, there are several measurements recorded at the same time for the same patient. In most cases, the values of these measurements are also identical. This suggests that some of the data are duplicated. This may occur if, for example, lab data are copied to the chart as chart data, if some chart itemid is copied to entries under a different itemid, or various other possible reasons. Here we investigate how frequently this occurs, and identify groups of itemids where such duplication is common.

# dups_equal <- dups %>% filter(ndistinct==1) %>% arrange(itemids)
# dups_unequal <- dups %>% filter(ndistinct > 1) %>% arrange(itemids)
dups_summary <- dups %>%
  group_by(itemids, nitemids) %>%
  summarize(ntot = n(),
            nequal = sum(ifelse(ndistinct==1, 1, 0)),
            nunequal = sum(ifelse(ndistinct > 1, 1, 0))
  ) %>%
  arrange(desc(nequal))
items <- c()
for (i in 1:nrow(dups_summary)) {
  items <- append(items, as.numeric(strsplit(as.character(dups_summary[i, "itemids"]), ", ")[[1]]))
}
item_names <- left_join(data.frame(itemid = unique(items)),
                        d_all,
                        by="itemid") %>%
  select(-desc) %>%
  arrange(itemid)
if (nrow(dups_summary) > 0) {
  kable(dups_summary, caption="Summary of identical data under different itemids")
}

itemids	nitemids	ntot	nequal	nunequal
51221, 813	2	229198	228692	506
51222, 814	2	171632	171567	65
50971, 1535	2	161085	160946	139
50983, 1536	2	152601	152524	77
50902, 1523	2	149106	149062	44
50960, 1532	2	146959	146923	36
50912, 1525	2	145734	145701	33
50931, 1529	2	145093	145036	57
51275, 1533	2	109798	109736	62
51274, 1286	2	105154	105124	30
50809, 1529	2	82307	82301	6
50912, 220615	2	46325	46308	17
51265, 227457	2	45663	45595	68
50960, 220635	2	44796	44779	17
51222, 220228	2	40358	40302	56
51275, 227466	2	30673	30611	62
50862, 1521	2	23194	23188	6
50867, 775	2	20270	20256	14
50802, 74	2	9382	9382	0
50821, 3837	2	9350	9350	0
50818, 3835	2	9329	9329	0
51200, 3754	2	8323	8317	6
51144, 3738	2	8022	8022	0
50862, 227456	2	7132	7130	2
224828, 776	2	5102	5067	35
50912, 220615, 1525	3	4694	4692	2
225624, 1162	2	4660	4659	1
50960, 220635, 1532	3	4607	4606	1
220546, 1542	2	4267	4250	17
51222, 220228, 814	3	4255	4236	19
225677, 1534	2	4143	4141	2
51275, 227466, 1533	3	3605	3586	19
227467, 1530	2	3224	3214	10
50956, 225672	2	3123	3123	0
225667, 816	2	2625	2615	10
50917, 801	2	2417	2417	0
50867, 220581	2	2292	2291	1
225668, 1531	2	1806	1795	11
51007, 1541	2	1690	1689	1
50863, 3728	2	1633	1633	0
50914, 792	2	1471	1471	0
51196, 1526	2	1225	1225	0
220587, 770	2	1106	1105	1
220644, 769	2	1105	1105	0
50909, 227463	2	1093	1066	27
50862, 227456, 1521	3	868	867	1
220632, 817	2	716	716	0
50811, 814	2	785	603	182
50966, 826	2	577	577	0
225642, 799	2	575	575	0
225639, 796	2	573	573	0
225640, 797	2	572	572	0
225641, 798	2	567	567	0
50867, 220581, 775	3	561	561	0
50971	1	562	477	85
50983	1	517	471	46
50902	1	486	457	29
50931	1	515	444	71
51221	1	493	419	74
227468, 1528	2	382	377	5
51007, 225695	2	367	367	0
50917, 227440	2	346	346	0
225638, 795	2	340	340	0
50912	1	369	335	34
51265	1	385	330	55
50820	1	298	295	3
50818	1	291	288	3
51222	1	350	288	62
50802	1	288	287	1
50821	1	289	286	3
220546	1	285	285	0
225624	1	281	281	0
225677	1	259	259	0
50960	1	270	256	14
50866, 220580	2	247	247	0
225674, 823	2	229	226	3
220228	1	217	217	0
227457	1	216	216	0
220615	1	211	211	0
220635	1	205	205	0
51274	1	215	192	23
225684, 836	2	181	180	1
50915, 225636	2	175	175	0
227467	1	160	160	0
224828	1	156	156	0
227466	1	121	121	0
225668	1	119	119	0
51275	1	147	118	29
50863	1	114	101	13
220587	1	98	98	0
220644	1	98	98	0
51146	1	104	97	7
51244	1	105	97	8
51254	1	105	97	8
51200	1	102	96	6
225693, 1540	2	90	90	0
50819, 505	2	99	90	9
50966, 227460	2	79	79	0
50922, 804	2	78	78	0
50809	1	76	76	0
225667	1	74	74	0
225639	1	73	73	0
225640	1	73	73	0
225641	1	73	73	0
225642	1	73	73	0
220632	1	71	71	0
51007, 225695, 1541	3	64	64	0
51491	1	79	64	15
220603, 1524	2	59	59	0
50811, 51222	2	698	57	641
50811, 220228	2	123	55	68
50862	1	57	52	5
50819	1	46	46	0
50811, 51222, 814	3	411	44	367
50826	1	43	42	1
225638	1	41	41	0
50922	1	43	41	2
227462, 805	2	40	40	0
227456	1	38	38	0
50917, 227440, 801	3	38	38	0
50811, 51222, 220228	3	225	35	190
50956	1	35	35	0
225672	1	30	30	0
220581	1	24	24	0
50867	1	24	21	3
220210, 618	2	135	20	115
50956, 820	2	21	20	1
51144	1	21	20	1
51196, 225636, 1526	3	19	19	0
224639	1	19	17	2
220045, 211	2	150	16	134
50809, 50931	2	347	16	331
50811	1	17	16	1
51143	1	17	16	1
51251	1	17	16	1
51255	1	16	16	0
220650, 1539	2	15	15	0
51196, 225636	2	14	14	0
227468	1	13	13	0
51493	1	18	13	5
51516	1	17	11	6
227470, 835	2	9	9	0
51094, 51491	2	24	9	15
225674	1	7	7	0
225684	1	7	7	0
225693	1	7	7	0
50809, 50931, 1529	3	136	7	129
50815, 223834	2	10	7	3
50866	1	8	7	1
211	1	17	6	11
5815	1	6	6	0
618	1	10	6	4
227463	1	5	5	0
50935	1	5	5	0
220228, 814	2	4	4	0
50811, 51222, 220228, 814	4	22	4	18
50931, 51478	2	1036	4	1032
8549	1	6	4	2
227462	1	3	3	0
227470	1	3	3	0
50815	1	3	3	0
50917	1	3	3	0
220603	1	2	2	0
220650	1	2	2	0
225695	1	2	2	0
50966, 227460, 826	3	2	2	0
51269	1	2	2	0
51478	1	2	2	0
51516, 1542	2	356	2	354
586	1	2	2	0
220615, 1525	2	1	1	0
220635, 1532	2	1	1	0
227440	1	1	1	0
50866, 774	2	1	1	0
51007	1	2	1	1
51094	1	1	1	0
51116, 51427	2	5	1	4
51125, 51436	2	5	1	4
51125, 51455	2	7	1	6
51196	1	1	1	0
51200, 51347, 3754	3	2	1	1
51352	1	27	1	26
51375, 51427	2	1	1	0
763	1	2	1	1
50809, 51478	2	13	0	13
50820, 51491	2	182	0	182
50909	1	8	0	8
50914	1	2	0	2
50931, 51478, 1529	3	77	0	77
50978	1	2	0	2
51110, 51440	2	1	0	1
51114, 51200	2	3	0	3
51114, 51444	2	2	0	2
51116	1	3	0	3
51116, 51244	2	9	0	9
51116, 51446	2	7	0	7
51117	1	2	0	2
51117, 51428	2	1	0	1
51120	1	3	0	3
51120, 51254	2	9	0	9
51125	1	2	0	2
51143, 51440	2	1	0	1
51144, 51344, 3738	3	1	0	1
51144, 51441	2	1	0	1
51200, 51347	2	1	0	1
51200, 51368	2	1	0	1
51244, 51375	2	1	0	1
51244, 51427	2	2	0	2
51244, 51446	2	8	0	8
51254, 51355	2	10	0	10
51254, 51379	2	1	0	1
51343	1	3	0	3
51344	1	5	0	5
51346	1	1	0	1
51347	1	6	0	6
51351	1	222	0	222
51351, 3770	2	6	0	6
51355	1	183	0	183
51358	1	15	0	15
51360	1	177	0	177
51360, 3791	2	6	0	6
51360, 51455	2	1	0	1
51361	1	1	0	1
51375	1	2	0	2
51376	1	1	0	1
51379	1	1	0	1
51382	1	2	0	2
51427	1	3	0	3
51428	1	2	0	2
51431	1	1	0	1
51436	1	3	0	3
51444	1	2	0	2
51446	1	3	0	3
51450	1	3	0	3
51450, 3779	2	1	0	1
51455	1	2	0	2
51455, 3791	2	1	0	1
51478, 1529	2	1	0	1
51493, 833	2	377	0	377
51516, 220546	2	156	0	156
51516, 220546, 1542	3	8	0	8

if (nrow(dups_summary) > 0) {
  kable(item_names, caption="Legend for duplicated data")
}

itemid	n	n_subjects	n_adm	label	category	linksto
74	9446	1343	1362	Base Excess	NA	chartevents
211	5179363	29898	34901	Heart Rate	NA	chartevents
505	350192	1833	1855	PEEP	NA	chartevents
586	2762	844	863	QTc	NA	chartevents
618	3384989	22312	27195	Respiratory Rate	NA	chartevents
763	47241	16409	19352	Daily Weight	NA	chartevents
769	41351	10660	12293	ALT	Enzymes	chartevents
770	41218	10622	12251	AST	Enzymes	chartevents
774	12	8	8	Ammonia	Chemistry	chartevents
775	21145	6868	7652	Amylase	Enzymes	chartevents
776	262053	15084	17229	Arterial Base Excess	ABG	chartevents
792	1489	165	204	Cyclosporin	Drug Level	chartevents
795	13582	6215	6916	Differential-Bands	Hematology	chartevents
796	21574	8916	10259	Differential-Basos	Hematology	chartevents
797	21575	8916	10260	Differential-Eos	Hematology	chartevents
798	21576	8916	10260	Differential-Lymphs	Hematology	chartevents
799	21575	8916	10260	Differential-Monos	Hematology	chartevents
801	2476	950	1034	Digoxin	Drug Level	chartevents
804	89	81	81	Ethanol	Drug Level	chartevents
805	1847	273	359	FK506	Drug Level	chartevents
813	274142	22405	27410	Hematocrit	Hematology	chartevents
814	187843	22188	27141	Hemoglobin	Hematology	chartevents
816	147034	13268	14630	Ionized Calcium	Chemistry	chartevents
817	23238	7990	9090	LDH	Enzymes	chartevents
820	77	63	63	Lipase	Enzymes	chartevents
823	42282	4725	4979	Mixed Venous O2% Sat	Blood Gases	chartevents
826	585	81	95	Phenobarbital	Drug Level	chartevents
833	175859	22167	27110	RBC	Hematology	chartevents
835	747	638	653	Sed Rate	Hematology	chartevents
836	8095	1768	1816	Serum Osmolality	Chemistry	chartevents
1162	151974	17295	21108	BUN	NA	chartevents
1286	106153	15700	18796	PT	NA	chartevents
1521	24310	8003	9159	Albumin	Chemistry	chartevents
1523	151252	17273	21078	Chloride	Chemistry	chartevents
1524	1794	1697	1711	Cholesterol	Chemistry	chartevents
1525	152641	17299	21113	Creatinine	Chemistry	chartevents
1526	1245	900	921	D-Dimer	Coags	chartevents
1528	15575	5560	5852	Fibrinogen	Coags	chartevents
1529	283798	17332	21147	Glucose	Chemistry	chartevents
1530	106359	15715	18817	INR	Coags	chartevents
1531	63233	9790	11149	Lactic Acid	Chemistry	chartevents
1532	153447	16941	20668	Magnesium	Chemistry	chartevents
1533	114460	15844	18935	PTT	Coags	chartevents
1534	134235	16954	20585	Phosphorous	Chemistry	chartevents
1535	242648	17346	21167	Potassium	Chemistry	chartevents
1536	172037	17306	21114	Sodium	Chemistry	chartevents
1539	802	697	712	Total Protein	Chemistry	chartevents
1540	2964	2371	2430	Triglyceride	Chemistry	chartevents
1541	1788	862	886	Uric Acid	Chemistry	chartevents
1542	138090	17247	21034	WBC	Hematology	chartevents
3728	1650	796	813	Alkaline Phosphatase	Chemistry	chartevents
3738	8049	5135	5180	Bands	Heme/Coag	chartevents
3754	8356	5315	5361	Eosinophils	Heme/Coag	chartevents
3770	8153	5188	5234	Lymphs	Heme/Coag	chartevents
3779	8153	5188	5234	Monos	CSF	chartevents
3791	8153	5188	5234	Polys	CSF	chartevents
3835	9450	1343	1362	pCO2	ABG'S	chartevents
3837	9450	1343	1362	pO2	ABG'S	chartevents
5815	1804455	14416	17386	HR Alarm [Low]	NA	chartevents
8549	1804988	14414	17384	HR Alarm [High]	NA	chartevents
50802	490527	31427	37357	Base Excess	Blood Gas	labevents
50809	196596	24029	26676	Glucose	Blood Gas	labevents
50811	89712	21698	23637	Hemoglobin	Blood Gas	labevents
50815	12307	7054	7449	O2 Flow	Blood Gas	labevents
50818	490504	31424	37348	pCO2	Blood Gas	labevents
50819	86911	12630	13831	PEEP	Blood Gas	labevents
50820	530657	32477	38918	pH	Blood Gas	labevents
50821	490522	31424	37351	pO2	Blood Gas	labevents
50826	83805	13738	15001	Tidal Volume	Blood Gas	labevents
50862	146652	28055	30997	Albumin	Chemistry	labevents
50863	207837	30060	33574	Alkaline Phosphatase	Chemistry	labevents
50866	3868	2143	2069	Ammonia	Chemistry	labevents
50867	63657	18362	18605	Amylase	Chemistry	labevents
50902	795412	41203	52853	Chloride	Chemistry	labevents
50909	13522	5429	5634	Cortisol	Chemistry	labevents
50912	797231	39349	50997	Creatinine	Chemistry	labevents
50914	8565	297	354	Cyclosporin	Chemistry	labevents
50915	1093	797	672	D-Dimer	Chemistry	labevents
50917	8152	2386	2620	Digoxin	Chemistry	labevents
50922	2626	1633	1671	Ethanol	Chemistry	labevents
50931	748836	38798	50396	Glucose	Chemistry	labevents
50935	10197	5916	5631	Haptoglobin	Chemistry	labevents
50956	65361	18636	20477	Lipase	Chemistry	labevents
50960	664079	38385	49763	Magnesium	Chemistry	labevents
50966	1767	190	225	Phenobarbital	Chemistry	labevents
50971	845365	41208	52877	Potassium	Chemistry	labevents
50978	3181	191	155	Rapamycin	Chemistry	labevents
50983	808328	41209	52860	Sodium	Chemistry	labevents
51007	18355	4261	3305	Uric Acid	Chemistry	labevents
51094	1157	838	566	pH	Chemistry	labevents
51110	160	132	114	Atypical Lymphocytes	Hematology	labevents
51114	566	376	325	Eosinophils	Hematology	labevents
51116	3346	1247	1306	Lymphocytes	Hematology	labevents
51117	2280	928	939	Macrophage	Hematology	labevents
51120	3345	1247	1306	Monocytes	Hematology	labevents
51125	3344	1247	1306	Polys	Hematology	labevents
51143	59334	19376	19935	Atypical Lymphocytes	Hematology	labevents
51144	75587	23119	24612	Bands	Hematology	labevents
51146	172093	37758	44247	Basophils	Hematology	labevents
51196	2877	1960	1565	D-Dimer	Hematology	labevents
51200	172095	37758	44247	Eosinophils	Hematology	labevents
51221	881653	45383	57103	Hematocrit	Hematology	labevents
51222	752277	45297	56950	Hemoglobin	Hematology	labevents
51244	172100	37759	44248	Lymphocytes	Hematology	labevents
51251	59206	19338	19902	Metamyelocytes	Hematology	labevents
51254	172099	37759	44248	Monocytes	Hematology	labevents
51255	59158	19331	19896	Myelocytes	Hematology	labevents
51265	778163	45303	56975	Platelet Count	Hematology	labevents
51269	2489	1228	1064	Promyelocytes	Hematology	labevents
51274	468914	37792	48226	PT	Hematology	labevents
51275	473449	37698	48064	PTT	Hematology	labevents
51343	310	268	235	Atypical Lymphocytes	Hematology	labevents
51344	260	241	218	Bands	Hematology	labevents
51346	16	10	8	Blasts	Hematology	labevents
51347	655	500	449	Eosinophils	Hematology	labevents
51351	4732	2840	2566	Lymphs	Hematology	labevents
51352	1138	869	785	Macrophage	Hematology	labevents
51355	4701	2840	2566	Monocytes	Hematology	labevents
51358	332	233	193	Other	Hematology	labevents
51360	4689	2840	2566	Polys	Hematology	labevents
51361	5	4	4	Promyelocytes	Hematology	labevents
51368	83	73	32	Eosinophils	Hematology	labevents
51375	683	468	320	Lymphocytes	Hematology	labevents
51376	236	192	124	Macrophage	Hematology	labevents
51379	683	468	320	Monocytes	Hematology	labevents
51382	683	468	320	Polys	Hematology	labevents
51427	1967	1529	1489	Lymphocytes	Hematology	labevents
51428	1078	912	877	Macrophage	Hematology	labevents
51431	1964	1529	1489	Monos	Hematology	labevents
51436	1967	1529	1489	Polys	Hematology	labevents
51440	159	150	124	Atypical Lymphocytes	Hematology	labevents
51441	121	115	107	Bands	Hematology	labevents
51444	778	627	507	Eosinophils	Hematology	labevents
51446	2613	1826	1638	Lymphocytes	Hematology	labevents
51450	2613	1826	1638	Monos	Hematology	labevents
51455	2612	1826	1638	Polys	Hematology	labevents
51478	11138	5714	5738	Glucose	Hematology	labevents
51491	116034	32430	37653	pH	Hematology	labevents
51493	43664	18708	18383	RBC	Hematology	labevents
51516	43097	18680	18310	WBC	Hematology	labevents
220045	2762225	17714	21924	Heart Rate	Routine Vital Signs	chartevents
220210	2737105	17707	21913	Respiratory Rate	Respiratory	chartevents
220228	137117	19253	23983	Hemoglobin	Labs	chartevents
220546	133356	19254	23983	WBC	Labs	chartevents
220580	694	503	521	Ammonia	Labs	chartevents
220581	8685	3497	3754	Amylase	Labs	chartevents
220587	37598	9775	11482	AST	Labs	chartevents
220603	1567	1484	1497	Cholesterol	Labs	chartevents
220615	153010	19313	24063	Creatinine	Labs	chartevents
220632	23125	7644	8789	LDH	Labs	chartevents
220635	147786	18788	23446	Magnesium	Labs	chartevents
220644	37625	9787	11494	ALT	Labs	chartevents
220650	785	670	694	Total Protein	Labs	chartevents
223834	202279	14550	17644	O2 Flow	Respiratory	chartevents
224639	46452	10152	11496	Daily Weight	General	chartevents
224828	140198	11601	13134	Arterial Base Excess	Labs	chartevents
225624	152475	19312	24062	BUN	Labs	chartevents
225636	764	535	542	D-Dimer	Labs	chartevents
225638	8610	3760	4135	Differential-Bands	Labs	chartevents
225639	20425	8774	10193	Differential-Basos	Labs	chartevents
225640	20424	8774	10193	Differential-Eos	Labs	chartevents
225641	20425	8774	10193	Differential-Lymphs	Labs	chartevents
225642	20425	8774	10193	Differential-Monos	Labs	chartevents
225667	80126	9894	10821	Ionized Calcium	Labs	chartevents
225668	69432	11936	13766	Lactic Acid	Labs	chartevents
225672	9766	4073	4401	Lipase	Labs	chartevents
225674	9867	2909	3024	Mixed Venous O2% Sat	Labs	chartevents
225677	135440	18061	22617	Phosphorous	Labs	chartevents
225684	6746	1578	1621	Serum Osmolality	Labs	chartevents
225693	3062	2367	2438	Triglyceride	Labs	chartevents
225695	1656	566	590	Uric Acid	Labs	chartevents
227440	1403	613	695	Digoxin	Labs	chartevents
227456	22945	8464	9673	Albumin	Labs	chartevents
227457	137454	19258	23984	Platelet Count	Labs	chartevents
227460	251	38	49	Phenobarbital	Labs	chartevents
227462	2159	283	407	FK506	Labs	chartevents
227463	2911	1528	1628	Cortisol	Labs	chartevents
227466	99769	16967	20645	PTT	Labs	chartevents
227467	93040	17060	20787	INR	Labs	chartevents
227468	14339	5496	5784	Fibrinogen	Labs	chartevents
227470	700	589	619	Sed Rate	Labs	chartevents

if (!is.null(dups)) {
  print(dups)
}

overlap_summary <- data.frame(table(dups_summary$nitemids))
names(overlap_summary) <- c("nitemids", "count")

There are 241 sets of items that share labels which contain some duplicated data (by our criterion: patient x charttime), amounting to 1985258 instances. Of these instances, the large majority appear to be entries for the same measured value at the same time under different itemids, i.e., duplicates. In fact, there are 1977973 instances where the value are the same, and 7285 where they differ. For sets such as 51221, 813, any analysis using these data should probably eliminate the duplicated values and merge those itemids if in fact their their overall value distributions look similar.

Among these sets, this is the distribution of the number of itemids in each set. If there are singletons, that means that some data are clearly duplicated within the same itemid.

if (!is.null(dups)) {
  dups_within_item <- dups %>% filter(nitemids==1) %>%
    group_by(itemids) %>%
    summarize(instances=n()) %>%
    arrange(desc(instances))
  
  if (nrow(dups_within_item) > 0) {
    print(dups_within_item)
  }
}

Finding related items whose labels don’t match exactly

Instead of grouping d_items by labels, as we did for d_bylabels, we form groups by choosing labels and then any items that contain that label as a subset of its string.

groups <- NULL
# for (l in d_items$label) {
#   related <- d_items %>% filter(grepl(l, label, fixed=TRUE))
#   if (nrow(related) > 1) {
#     groups[[l]] <- related
#     print(related)
#   }
# }
sid <- function(ids) {
  l <- as.character((d_all %>% filter(itemid %in% ids))[1, "desc"])
  dat <-  all %>% 
    filter(itemid %in% ids) %>% 
    left_join(d_all, by="itemid")
  binw <- bin_width(dat$valuenum)
  if (nrow(table(dat$valuenum)) <=2) {
    print(sprintf("%s is binary", l))
  } else {
    dat  %>%
      rid_outliers() %>%
      ggplot(aes(valuenum, fill=desc)) + ggtitle(l) + geom_histogram(binwidth = binw)
  }
}
sid(227292)

[1] "227292-mv (182070) Arterial Line Dressing Occlusive {Access Lines - Invasive} is binary"

sid(51176)

sid(51132)

sid(29)

sid(224149)

sid(50856)

Although the above is helpful for exploration, it is not a good systematic basis. For example, a string such as “TP” appears in too many other unrelated labels.

Manually guided exploration

The analysis above is based on the assumption that it is most helpful to compare distributions of variables that have the same label. However, we know that the same label does not always mean that the values should be expected to be the same (e.g., the same substance being measured in different body fluids), and, conversely, that sometimes different labels represent the same measurement but expressed differently (e.g., various way to say arterial blood pressure). The latter is described on GitHub at phys_acuity_modelling/resources/itemid_to_variable_map.csv in what is so far a private repository. I have also copied that file to the current R project directory, but of course that version will not be updated automatically from changes made on GitHub.

itemid_to_variable <- read.csv("itemid_to_variable_map.csv") %>% filter(LEVEL2 != "")

compare_distributions <- function(itemids, label="") {
  
  dat <- all %>% filter(itemid %in% itemids) %>%
    rid_outliers(outliers_fraction) %>%
    left_join(d_all, by="itemid")
  
  if (nrow(dat) == 0) {
    print(sprintf("No data for %s", paste(itemids, collapse=", ")))
  } else {
    
    if (label=="") {
      label <- dat$label[1]
    }
     
    print(sprintf("%d rows for '%s' (%s)", nrow(dat), label, paste(itemids, collapse=",")))
 
    binw <- bin_width(dat$valuenum)
    
    print(dat %>%
            ggplot(aes(valuenum, fill=desc)) + 
            geom_histogram(binwidth = binw) +
            facet_wrap(~ desc) +
            theme(legend.position="none") +
            ggtitle(label))
    print(dat %>%
            ggplot(aes(valuenum, fill=desc)) +
            geom_histogram(binwidth = binw, alpha=0.5) +
            ggtitle(label) +
            theme(legend.position="bottom", legend.direction="vertical"))
    print(dat %>%
            ggplot(aes(x=desc, y=valuenum, color=desc)) +
            geom_violin() +
            coord_flip() +
            theme(legend.position="none") +
            ggtitle(label))
    print(dat %>%
            ggplot(aes(valuenum, color=desc)) +
            stat_ecdf(aes(linetype = desc)) +
            theme(legend.position="bottom", legend.direction="vertical") +
            ggtitle(sprintf("CDF of %s", label)))
    
    # Collect duplication data while producing plots
    if (check_for_duplicates) {
      mults <- dat %>%
        group_by(subject_id, charttime) %>%
        summarize(
          itemids = paste(unique(itemid), collapse=", "),
          nitemids = n_distinct(itemid),
          nvals = n(),
          ndistinct = n_distinct(valuenum),
          vals = paste(valuenum, collapse=", "),
          val_mean = mean(valuenum),
          val_low = min(valuenum),
          val_high = max(valuenum)
        ) %>%
        ungroup() %>%
        filter(nvals > 1)
      if (nrow(mults) > 0) {
        # print(sprintf("%s has %d duplicate rows:", label, nrow(mults)))
        # print(table(mults$ndistinct))
        smry <- mults %>% group_by(itemids) %>%
          summarize(n = n(),
                    nsame = sum(ifelse(ndistinct==1, 1, 0)),
                    ndiff = sum(ifelse(ndistinct==1, 0, 1)))
        kable(smry, caption="Duplicate rows")
      }
    }
  }
}

We examine each “Level 2” concept from this list in turn:

Alanine aminotransferase (ALT)

compare_distributions(c(50861,769,220644))

[1] "295756 rows for 'Alanine Aminotransferase (ALT)' (50861,769,220644)"

itemids	n	nsame	ndiff
220644	82	82	0
50861	98	83	15
50861, 220644	11728	11724	4
50861, 220644, 769	1104	1104	0
50861, 769	39545	39531	14

Albumin

compare_distributions(c(50862,772,1521,227456,3727))

[1] "223304 rows for 'Albumin' (50862,772,1521,227456,3727)"

itemids	n	nsame	ndiff
227456	38	38	0
50862	57	52	5
50862, 227456	7124	7122	2
50862, 227456, 772, 1521	868	867	1
50862, 3727	542	542	0
50862, 772	6410	6410	0
50862, 772, 1521	23185	23179	6
772, 1521	65	65	0

Alkaline Phosphatase

compare_distributions(c(50863,773,225612,3728))

[1] "283868 rows for 'Alkaline Phosphatase' (50863,773,225612,3728)"

itemids	n	nsame	ndiff
225612	79	79	0
50863	92	80	12
50863, 225612	11453	11450	3
50863, 225612, 773	1053	1052	1
50863, 3728	1631	1631	0
50863, 773	38804	38789	15

Sysematically explore the groupings of variables from itemid_to_variable_map.csv.

# library(pander)
# library(purrr)
vargroups <- itemid_to_variable %>% split(.$LEVEL2)
for (i in seq_along(vargroups)) {
  g <- vargroups[[i]]
  if (nrow(g) > 0) {
    # pandoc.header(list(as.character(g$LEVEL2[1])))
    print(g$LEVEL2[1])
    compare_distributions(g$ITEMID)
  }
}

[1] Alanine aminotransferase
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "295756 rows for 'Alanine Aminotransferase (ALT)' (50861,769,220644)"
[1] Albumin
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "227690 rows for 'Albumin' (50862,772,1521,227456,3727,50835,51025,51046,51069)"
[1] Alkaline phosphate
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "283868 rows for 'Alkaline Phosphatase' (50863,773,225612,3728)"
[1] Anion gap
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "930089 rows for 'Anion Gap' (50868,227073,3732)"
[1] Asparate aminotransferase
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "295943 rows for 'Asparate Aminotransferase (AST)' (50878,770,220587)"
[1] Basophils
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "171877 rows for 'Basophils' (51146,51442,51345,51112,51387,51367)"
[1] Bicarbonate
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "935150 rows for 'Bicarbonate' (50882,227443,51076,50803)"
[1] Bilirubin
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "499067 rows for 'Bilirubin, Direct' (51028,50883,803,225651,51012,50838,51049,50885,1538,848,225690,50884,3765,51464,51465)"
[1] Blood culture
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1 rows for 'Blood Cultures' (3333,50886)"
[1] Blood urea nitrogen
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1281200 rows for 'Urea Nitrogen' (51006,781,1162,225624,3737)"
[1] Calcium
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1254630 rows for 'Calcium, Total' (50808,786,1522,3746,51066,51029,50893,51077,225625)"
[1] Calcium ionized
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "224904 rows for 'Ionized Calcium' (816,225667)"
[1] Calcium Ionized
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "93 rows for 'Ion Calcium' (3766)"
[1] Capillary refill rate
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "No data for 3348, 115, 8377"
[1] Chloride
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1373552 rows for 'Chloride' (788,1523,3747,4193,50839,51030,51013,51050,220602,51062,51078,226536,50902,50806)"
[1] Cholesterol
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "61506 rows for 'Cholesterol, HDL' (789,1524,220603,3748,50840,51031,50904,50905,50906,51051,50907)"
[1] CO2 (ETCO2, PCO2, etc.)
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "954269 rows for 'Calculated Total CO2' (50804,777,225698,3810,3808,857,3809,4199,223679,226062,3773,3830,3832)"
[1] Creatinine
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1328144 rows for 'Creatinine' (791,1525,220615,3750,51067,50841,51032,51052,51081,51082,51106,51080,50912)"
[1] Diastolic blood pressure
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "6301729 rows for 'Arterial Blood Pressure diastolic' (8368,220051,225310,8555,8441,220180,8502,8440,8503,8504,8507,8506,224643)"
[1] Eosinophils
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "10985 rows for 'Eosinophils' (3754,51474,51444,51347,51419,51114,51368)"
[1] Fraction inspired oxygen
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1469730 rows for 'Inspired O2 Fraction' (3420,223835,3422,189,727)"
[1] Glascow coma scale eye opening
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1521073 rows for 'Eye Opening' (184,220739)"
[1] Glascow coma scale motor response
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1514681 rows for 'Motor Response' (454,223901)"
[1] Glascow coma scale total
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "945427 rows for 'GCS Total' (198)"
[1] Glascow coma scale verbal response
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1516822 rows for 'Verbal Response' (723,223900)"
[1] Glucose
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "2497714 rows for 'Glucose' (50931,807,811,1529,50809,51478,3745,225664,220621,226537)"
[1] Heart Rate
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "7868558 rows for 'Heart Rate' (211,220045)"
[1] Height
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "23864 rows for 'Height' (226707,226730,1394)"
[1] Hematocrit
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1410213 rows for 'Hematocrit' (813,3761,220545,51221,50810)"
[1] Hemoglobin
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1172056 rows for 'Hemoglobin' (814,220228,50852,50855,51222,50811)"
[1] INR
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "467604 rows for 'INR(PT)' (51237)"
[1] Lactate
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "185828 rows for 'Lactate' (50813)"
[1] Lactate dehydrogenase
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "108530 rows for 'Lactate Dehydrogenase (LD)' (50954,51054)"
[1] Lactic acid
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "204302 rows for 'Lactic Acid' (818,225668,1531)"
[1] Lymphocytes
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "244286 rows for 'Atypical Lymphocytes' (51244,51116,51446,51427,51375,51133,51143,51343,51110,51440,51385,51365,51245)"
[1] Magnesium
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1151841 rows for 'Magnesium' (821,1532,220635,50960)"
[1] Mean blood pressure
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "6298103 rows for 'Arterial Blood Pressure mean' (52,220052,225312,224,6702,224322,456,220181,3312,3314,3316,3322,3320)"
[1] Mean corpuscular hemoglobin
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "740209 rows for 'MCH' (51248)"
[1] Mean corpuscular hemoglobin concentration
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "741254 rows for 'MCHC' (51249)"
[1] Mean corpuscular volume
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "741200 rows for 'MCV' (51250)"
[1] Monocytes
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "175943 rows for 'Monocytes' (51254,51355)"
[1] Neutrophils
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "171274 rows for 'Neutrophils' (51256)"
[1] Oxygen saturation
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "7976868 rows for 'Oxygen Saturation' (834,50817,8498,220227,646,220277)"
[1] Partial pressure of carbon dioxide
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "923584 rows for 'pCO2' (50818,3835,3836,3784,778,220235,4201)"
[1] Partial pressure of oxygen
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "270215 rows for 'Arterial PaO2' (779,490)"
[1] Partial thromboplastin time
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "829240 rows for 'PTT' (825,1533,227466,51275)"
[1] Peak inspiratory pressure
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "149193 rows for 'PIP' (507)"
[1] pH
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1387343 rows for 'pH' (50820,51491,3839,1673,50831,51094,780,1126,223830,4753,4202,860,220274)"
[1] Phosphate
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "585016 rows for 'Phosphate' (50970)"
[1] Platelets
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1104267 rows for 'Platelet Count' (51265,828,227457,3789)"
[1] Positive end-expiratory pressure
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "478153 rows for 'PEEP' (505,50819,224700)"
[1] Potassium
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1815358 rows for 'Potassium' (829,1535,3792,227442,227464,50971,50822)"
[1] Prothrombin time
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1127001 rows for 'PT' (815,824,1530,1286,227467,227465,51274)"
[1] Pupillary response left
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "No data for 8466, 3593"
[1] Pupillary response right
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "No data for 584, 8530"
[1] Pupillary size left
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "No data for 8467"
[1] Pupillary size right
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "No data for 585"
[1] Red blood cell count
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "993516 rows for 'Red Blood Cells' (51279,833,51493,3799,4197,51127,51362,51278,51383,51438,51457)"
[1] Respiratory rate
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "9294868 rows for 'Respiratory Rate' (618,220210,3603,224689,614,651,224422,615,224690)"
[1] Sodium
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1455574 rows for 'Sodium' (837,1536,3803,220645,226534,50983,50824)"
[1] Systolic blood pressure
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "6306342 rows for 'Arterial Blood Pressure systolic' (51,220050,225309,6701,455,220179,3313,3315,442,3317,3323,3321,224167,227243)"
[1] Temperature
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "3386638 rows for 'Temperature Fahrenheit' (3655,677,676,223762,3654,678,223761,679)"
[1] Troponin-I
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "8277 rows for 'Troponin I' (51002,851)"
[1] Troponin-T
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "85471 rows for 'Troponin T' (51003,227429)"
[1] Urine Appearance
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "No data for 8480"
[1] Urine Color
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "No data for 706"
[1] Urine output
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "No data for 40055, 40056, 40057, 40061, 40065, 40069, 40085, 40086, 40094, 40096, 40405, 40428, 40473, 40651, 40715, 43175, 226557, 226558, 226559, 226560, 226561, 226563, 226564, 226565, 226567, 226584, 226627, 227510"
[1] Weight
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "4531802 rows for 'Admission Weight (Kg)' (763,224639,226512,3580,3693,3581,226531,3582,581,3723,3583,3692,580,733,225124,4183,226846)"
[1] White blood cell count
70 Levels:  Alanine aminotransferase Albumin Alkaline phosphate Anion gap Asparate aminotransferase Basophils Bicarbonate Bilirubin Blood culture ... White blood cell count
[1] "1418537 rows for 'White Blood Cells' (861,1127,1542,220546,51516,4200,51301,51300)"