Concept objects are used in ricu as a way to specify how a clinical concept, such as heart rate can be loaded from a data source and are mainly consumed by load_concepts(). Several functions are available for constructing concept (and related auxiliary) objects either from code or by parsing a JSON formatted concept dictionary using load_dictionary().

  description = name,
  category = NA_character_,
  aggregate = NULL,
  target = "ts_tbl",
  class = "num_cncpt"


init_cncpt(x, ...)

# S3 method for num_cncpt
init_cncpt(x, unit = NULL, min = NULL, max = NULL, ...)

# S3 method for unt_cncpt
init_cncpt(x, unit = NULL, min = NULL, max = NULL, ...)

# S3 method for fct_cncpt
init_cncpt(x, levels, ...)

# S3 method for cncpt
init_cncpt(x, ...)

# S3 method for rec_cncpt
  callback = paste0("rename_data_var('", x[["name"]], "')"),
  interval = NULL,







The name of the concept


Zero or more itm objects


String-valued concept description


String-valued category


NULL or a string denoting a function used to aggregate per id and if applicable per time step


Further specification of the cncpt object (passed to init_cncpt())


The target object yielded by loading


NULL or a string-valued sub-class name used for customizing concept behavior


Object to query/dispatch on


A string, specifying the measurement unit of the concept (can be NULL)

min, max

Scalar valued; defines a range of plausible values for a numeric concept


A vector of possible values a categorical concept may take on


Name of a function to be called on the returned data used for data cleanup operations


Time interval used for data loading; if NULL, the respective interval passed as argument to load_concepts() is taken


Constructors and coercion functions return cncpt and concept objects, while inheritance tester functions return logical flags.


In order to allow for a large degree of flexibility (and extensibility), which is much needed owing to considerable heterogeneity presented by different data sources, several nested S3 classes are involved in representing a concept. An outline of this hierarchy can be described as

  • concept: contains many cncpt objects (of potentially differing sub-types), each comprising of some meta-data and an item object

  • item: contains many itm objects (of potentially differing sub-types), each encoding how to retrieve a data item.

The design choice for wrapping a vector of cncpt objects with a container class concept is motivated by the requirement of having several different sub-types of cncpt objects (all inheriting from the parent type cncpt), while retaining control over how this homogeneous w.r.t. parent type, but heterogeneous w.r.t. sub-type vector of objects behaves in terms of S3 generic functions.

Each individual cncpt object contains the following information: a string- valued name, an item vector containing itm objects, a string-valued description (can be missing), a string-valued category designation (can be missing), a character vector-valued specification for an aggregation function and a target class specification (e.g. id_tbl or ts_tbl). Additionally, a sub- class to cncpt has to be specified, each representing a different data-scenario and holding further class-specific information. The following sub-classes to cncpt are available:

  • num_cncpt: The most widely used concept type is indented for concepts representing numerical measurements. Additional information that can be specified includes a string-valued unit specification, alongside a plausible range which can be used during data loading.

  • fct_cncpt: In case of categorical concepts, such as sex, a set of factor levels can be specified, against which the loaded data is checked.

  • lgl_cncpt: A special case of fct_cncpt, this allows only for logical values (TRUE, FALSE and NA).

  • rec_cncpt: More involved concepts, such as a SOFA score can pull in other concepts. Recursive concepts can build on other recursive concepts up to arbitrary recursion depth. Owing to the more complicated nature of such concepts, a callback function can be specified which is used in data loading for concept-specific post- processing steps.

  • unt_cncpt: A recent (experimental) addition which inherits from num_cncpt but instead of manual unit conversion, leverages

Class instantiation is organized in the same fashion as for item objects: concept() maps vector-valued arguments to new_cncpt(), which internally calls the S3 generic function init_cncpt(), while new_concept() instantiates a concept object from a list of cncpt objects (created by calls to new_cncpt()). Coercion is only possible from list and cncpt, by calling as_concept() and inheritance can be checked using is_concept() or is_cncpt().


if (require(mimic.demo)) { gluc <- concept("glu", item("mimic_demo", "labevents", "itemid", list(c(50809L, 50931L))), description = "glucose", category = "chemistry", unit = "mg/dL", min = 0, max = 1000 ) is_concept(gluc) identical(gluc, load_dictionary("mimic_demo", "glu")) gl1 <- new_cncpt("glu", item("mimic_demo", "labevents", "itemid", list(c(50809L, 50931L))), description = "glucose" ) is_cncpt(gl1) is_concept(gl1) conc <- concept(c("glu", "lact"), list( item("mimic_demo", "labevents", "itemid", list(c(50809L, 50931L))), item("mimic_demo", "labevents", "itemid", 50813L) ), description = c("glucose", "lactate") ) conc identical(as_concept(gl1), conc[1L]) }
#> [1] FALSE