Skip to content

Screenshot

Metrics Definitions

Flask enables you to create patient-level metrics to measure study performance and progress.

Metrics can be computed using a pattern language referencing data variables in your study.

The Metrics pattern language

The pattern language was designed to enable you to create metrics of performance, exceptions, missing data, or critical values on the patient timeline.

The language uses item variable names - like $CHANGE_IN_SEVERITY.

Simple single variable expressions

You can use a simple expressions like $CHANGE_IN_SEVERITY == '0',

Count things

You can count things using COUNT and FILTER count($CHANGE_IN_SEVERITY ) == '0' or filter($PATIENT_REPORTED_RESTLESSNESS) == '0'.

Compound expressions

You can combine expressions using logical operators like &&(AND) or ||(OR). The logical operator !(NOT) can also be used.

Examples

  • $ELIGIBILITY == '1' && count($TREATMENT_SESSIONS) > '0'
  • $CHANGE_IN_SEVERITY > 2 || $PATIENT_REPORTED_RESTLESSNESS == '0'
  • count($CONMED) <= '5'

Functions

Counting things

count(variable, period)- the count function counts a variable entered in a time period.

The count function will return the number of records in the dataset where the variable has any value. If the count is 0 then the data is missing.

The count function takes 2 arguments:

  • variable - the unique variable name of the CRF item.
  • period - an optional time period counting data for each patient.

Period is a time interval like 24 hours

!!info

You can use the following time units:

* milliseconds
* seconds
* minutes
* hours
* days
* weeks
* months
* years

Usage examples:

  • count($ADVERSE_EVENTS) - Count adverse events for each patients.
  • count($ADVERSE_EVENTS, '24 hours') - Count adverse events for patients in the last 24 hours.
  • count($ADVERSE_EVENTS, '7 days') - Count for the last 7 days
  • Say you want to find patients missing blood test data in the last 24 hours - you'd do this: 
count($BLOOD_TEST, '24 hours') == '0';

The metric will have the value of 1 for all patients missing blood tests in the last 24 hours.

Filter - more complex counting

filter(variable, period, value, take) - the filter function enables you to filter on values looking forwards and backwards on the patient journey.

Like count, filter returns the number of records entered into the database for the variable.

Arguments:

  • variable - the unique variable of the CRF item
  • period - an optional time period for filtering patient data according to CRF create date.
  • value - an optional filter using a comparison operator like >2, yes, ==5
  • take - an optional argument to take the first N records or last N records in the dataset for each patient.

Info

You can use the following operators:

  • ==
  • !=
  • >=
  • <=
  • >
  • <

Example

  • ==2
  • some string value
  • >=5

the expression ==2 and 2 or ==yes and yes are equivalent!

The value of the take parameter will be a string like '-1', '1', '5'

The take parameter enables you to take events from the beginning or end of the patient timeline:

Examples

  • '1' - The first event
  • '2' - The first 2 events
  • '-1' - The last event
  • '-4' - The last 4 events

the records returned from the database are ordered by date created.

Filter arguments are evaluated in this order:

  1. filtering by variable
  2. filtering by period
  3. filtering by 'take'
  4. filtering by 'value'

if you don't need to use some parameter you can set null instead.

Usage examples

  • filter($SCREENED) - count the $SCREENED variable for each patient.
  • filter($EPRO_DIARY, '24 hours') - count 1 if any diaries were entered by each patient in the last 24 hours
  • filter($TOOK_ANALGESICS, '24 hours', '>2') - count 1 for each patient who took pain analgesics more than twice in the past 24 hours
  • filter($TOOK_ANALGESICS, null, '>2') - count 1 for each patient who took pain analgesics more than twice in the study
  • filter($TOOK_ANALGESICS, null, '>2', '-2') - count 1 for each patient who took pain analgesics more than twice in the last 2 study events

You can do a comparison operation on the filter count and return a 0/false or 1/true for the metric:

    filter($TOOK_ANALGESICS, '24 hours', '>2') > '0';

Or just do the filter count and return a number for the metric like this:

filter($ELSE_ONE_VARIABLE, '24 hours', '>2')

Under the hood of the pattern language for the curious

You can use a simple expressions like $AGE == '77'.

In this case this expression will be replaced to filter($AGE, null, '== 77') != 0

Example

Input Output
$SOME_VARIABLE == '0' filter($SOME_VARIABLE, null, '== 0') != 0
$SOME_VARIABLE filter($SOME_VARIABLE, null, null) != 0
$SOME_VARIABLE == '3'&& $ELSE_ONE_VARIABLE > '2' filter($SOME_VARIABLE, null, '== 3') != 0 && filter($ELSE_ONE_VARIABLE, null, '>2') != 0