Description of column-types used to define controls and events

EVENT ID (formerly EVID): event identification data item.

The EVENT ID column permits to assign lines to be dose or response-lines, and to define washouts/resets. The column is not mandatory, as dose and response lines can be recognized based on the content (see here for details). The EVENT ID column can take 5 different values:

  • EVENT ID = 0: observation event, the line is a response-line. The dose-related informations (AMOUNT column, etc) are ignored.
  • EVENT ID = 1: dose event, the line is a dose-line. The response-related informations (OBSERVATION column, etc) are ignored.
  • EVENT ID = 2: other event. Both the dose information and the observation are ignored. Note that no prediction will be outputted for that time in the output files.
  • EVENT ID = 3: reset + dose event. The system is reset to the initial values and the dose is applied immediately after. To do a reset without applying a new dose, the dose amount can be set to zero. Unlike EVENT ID = 4, no occasions are created. The line is a dose-line (i.e the response-related informations are ignored).
  • EVENT ID = 4: reset + dose event. The system is reset to the initial values and the dose is applied immediately after. If at least one EVENT ID = 4 appear at a position which is not the first record of an individual, a OCCASION column-type called OCCevid will be created. It may (but doesn’t have to) be used to define inter-occasion variability. To do a reset without applying a new dose, the dose amount can be set to zero. The line is a dose-line (i.e the response-related informations are ignored).

All combinations of EVENT ID and IGNORED OBSERVATION values are possible and EVENT ID has priority over IGNORED OBSERVATION (for instance if EVENT ID = 1 and IGNORED OBSERVATION = 0, the line is a dose-line). For all EVENT ID values, the REGRESSOR values are taken into account.

Examples:

  • Example with EVENT ID = 0, 1, 2 and 3:

  • Example with EVENT ID = 4:

Format restrictions:

  • A data set shall not contain more than one column with column-type EVENT ID.
  • EVENT ID shall contain an integer in {0, 1, 2, 3, 4}.
  • When a line is tagged EVENT ID = 0, and IGNORED OBSERVATION = 0, the value contained in column OBSERVATION shall be a double.
  • When a line is tagged EVENT ID = 1, 3 or 4, the value in the AMOUNT column shall be a double.

 

IGNORED OBSERVATION (formerly MDV): ignores the observations (missing dependent variable).

The IGNORED OBSERVATION column-type enables to tag lines for which the information in the OBSERVATION column-type should be ignored (because they are outliers, or because one wants to ignore all PK measurements for instance). The column is not mandatory. Several IGNORED OBSERVATION columns are possible (see below) and 3 values are possible:

  • IGNORED OBSERVATION = 0: the value in the OBSERVATION column is not ignored. If in addition EVENT ID = 0, the value in the OBSERVATION column has to be a double.
  • IGNORED OBSERVATION = 1: the value in the OBSERVATION column is ignored.
  • IGNORED OBSERVATION = 2: identical to IGNORED OBSERVATION = 1. Note that no prediction will be made at that time point in the output files.

If there are both an IGNORED OBSERVATION and an EVENT ID column, the EVENT ID column has priority to define dose and response-lines (see here for more details). It is not necessary to set IGNORED OBSERVATION = 1 when EVENT ID = 1.

For all IGNORED OBSERVATION values, the REGRESSOR values are taken into account.

It is possible to have multiple IGNORED OBSERVATION columns in order to ignore observations for different reasons (for instance an “outliers” column to ignore outliers and a “placebo” column to ignore individuals which received only placebo, both tagged as IGNORED OBSERVATION column-type). When there are multiple IGNORED OBSERVATION columns, a synthetic value is computed as:

  • if IGNORED OBSERVATION = 0 in all columns, then resulting synthetic IGNORED OBSERVATION equals 0.
  • if IGNORED OBSERVATION = 1 or 2 in at least one column, then the resulting synthetic IGNORED OBSERVATION equals 1.

Examples:

  • Example with IGNORED OBSERVATION = 0, 1 or 2:

Format restrictions:

  • IGNORED OBSERVATION shall contain only integers in {0, 1, 2}.
  • If IGNORED OBSERVATION = 0 and EVENT ID = 0, the value in the OBSERVATION column has to be a double.

IGNORED LINE: ignores all the element of the line

Starting from the 2019 version, the IGNORED LINE column-type enables to tag lines for which all the information should be ignored (because they are outliers, or because one wants to ignore all PK measurements for instance). Contrary to the IGNORED OBSERVATION column-type that only ignores the observation value, the IGNORED LINE column-type allows to ignore doses and regressor values in addition to the observation values.
The column is not mandatory.

Starting from the 2021 version, there can be several columns IGNORED LINE, while in previous versions there can be only one such column.

Starting from the 2024 version, IGNORED LINE column can contain integers larger than 1, while previously the values were restricted to 0 and 1.

Format restrictions:

  • IGNORED LINE shall contain only non-negative integers.