How do I build scoring functions into scorecard objectives?

ProKnow DS allows you to build scoring functions into the computed metric objectives for patient scorecards. If there is at least one scoring function defined within a scorecard, the total score and maximum possible points across all scoring functions within the scorecard will be displayed in the toolbar at the top.

2019-07-24_08-55-26.png

There are two different configurations for objectives that support scoring functions. The first is the Pass/Fail configuration, and the second the Performance Bin configuration. For each configuration, this article will explain how to define a valid scoring configuration and how the scores will be computed.

Pass/Fail

A Pass/Fail scoring configuration consists of two objectives. One objective must be labeled PASS [N1] (case insensitive), where N1 is any number, and the other must be labeled FAIL (case insensitive) or FAIL [N2] (case insensitive), where N2 is any number.

The resulting score function will be a step function with N1 points assigned to any metric value that is "better" than the passing threshold and N2 points (or 0 if N2 is omitted) to any metric value is "worse" than the passing threshold. Please note that in this case "better" and "worse" are based on the direction of the pass and fail objective levels. In the rare case that a metric value is exactly the threshold score, the inclusion marker will be used to determine the assigned score.

Example 1

2019-07-24_08-49-23.png

In this Pass/Fail example, any value less than 66.5 will be assigned the default score of 0, whereas any value greater than or equal to 66.5 will be assigned the score of 1.

1.jpg

Example 2

2019-07-24_08-50-33.png

In this Pass/Fail example, any value greater than 48 will be assigned the default score of 0, whereas any value less than or equal to 48 will be assigned the score of 1.

2.jpg

Performance Bin

A Performance Bin configuration consists of at least two objectives and up to five objectives. The objective labels must begin with one of the following (case insensitive): UNACCEPTABLE, MARGINAL, ACCEPTABLE, GOOD, IDEAL. All levels other than UNACCEPTABLE must specify a score in brackets next to the label like this: IDEAL [3]. UNACCEPTABLE may also include a score in brackets but will default to 0 if not provided. The points provided in brackets must be in either non-decreasing or non-increasing order. Labels must also be in logical order (the order listed above, forward or backward). Some examples of valid and invalid objective orderings are listed below:

  • Valid: UNACCEPTABLE, ACCEPTABLE, IDEAL
  • Valid: IDEAL, GOOD, ACCEPTABLE, MARGINAL, UNACCEPTABLE
  • Invalid: IDEAL, ACCEPTABLE, GOOD, MARGINAL, UNACCEPTABLE
  • Invalid: UNACCEPTABLE, IDEAL, MARGINAL

To determine the resulting score function, we start from the level that represents the "worst" outcome (usually UNACCEPTABLE). Any metric value located in this range is assigned the score associated with this level. Then as we move toward better outcomes, the score will be linearly interpolated between the 2 nearest levels. If the metric value is better than the highest threshold level, the maximum score defined by the highest objective will be assigned.

Example 1

2019-07-24_08-46-59.png

In this performance bin example, any value less than 90 will be assigned a score of -10. A score of 0 will be assigned for a metric value equal to 90. As the metric value increases, the score will linearly increase from 0 to 1 between 90 and 93, from 1 to 2 between 93 and 95, and from 2 to 3 between 95 and 97. Any metric value greater than 97 will be assigned a score of 3.

3.jpg

Example 2

2019-07-24_08-48-12.png

In this performance bin example, any value greater than 48 will be assigned a score of 0 (the default for UNACCEPTABLE). A score of 0 will also be assigned for a metric value equal to 48. As the metric value decreases, the score will linearly increase from 0 to 1 between 48 and 45, from 1 to 2 between 45 and 40, and from 2 to 3 between 40 and 30. Any metric value less than 30 will be assigned a score of 3.

4.jpg

Was this article helpful?
0 out of 0 found this helpful
Have more questions? Submit a request

Comments

0 comments

Article is closed for comments.