Use Z-scores to Interpret Wellness Data

Context

Wellness workflows are generally implemented because we are interested in the wellbeing of our team. Tracking wellness is more than just knowing a person' daily responses though- often it is useful to understand their responses over a period time. As a result, a more powerful way to monitor wellbeing can be to highlight relative changes to a person’s state as opposed to interpreting wellbeing based on a standard threshold. For example, Person A may consistently report that they are feeling very fresh while Person B may report that they usually feel mildly fatigued. If responses are flagged whenever someone reports feeling mildly tired, Person B will often be flagged even though this is their usual state.

A simple way to calculate the relative change is by calculating z-scores. A z-score calculates how many standard deviations a score is from its mean. Positive z-scores indicate that the score is higher than the mean while negative z-scores indicate that the score is less than the mean. Using our example from before, if Person B usually reports mild fatigue but wakes up feeling very refreshed, we would expect to see a positive z-score. If Person A usually feels very refreshed but wakes up feeling somewhat fatigued, we would expect to see a negative z-score.

Goals

This tutorial will take you through the process of setting up the necessary calculations in an event form, reporting the data and providing feedback mechanisms.

Outcomes

1. Calculate z-scores using history calculations, option calculations and simple calculations within an event form.
2. Apply performance standards to z-scores.
3. Visualize the z-score data with tile and table widgets using the dashboard builder.
4. Set up performance alerts to provide feedback based on responses.

Outcome 1: Calculate z-scores within an event form

1. Create slider, single select or dropdown fields for each of your wellness metrics. Assign a unique score to each option. Click the Update button after adding new options and scores to save the values.

   

2. Add a calculation for each wellness metric field to return the score of the option selected.

   

3. Use history calculations to calculate the mean and standard deviation for each metric. Reference the calculation fields created in step 2.

   

4. Using a calculation field, reference the calculations set up in steps 2 and 3 to calculate the z-score for each metric. The format for calculating a z-score in AMS is:

if(safe(Standard deviation)=0, 0, 
((Current score - Mean) / Standard deviation))

What is the safe() function? The safe function is useful for when a calculation involves a number, but the number is not always entered. Since a calculation like division can only work if the denominator is greater than 0, safe will provide either the number, or 0. For example, if there is no value in the standard deviation field then Fatigue SD will be empty and the Z score calculation will fail. In this case, safe(Fatigue SD) = 0.

5. Use option calculations to convert the z-score value to a descriptive word or phrase. Not everyone will be familiar with z-scores, so this step will assist with interpreting the data. The calculation we have used to convert z-scores to a descriptive phrase is:

if(isempty(Z score), "NA",
if(Z-score > threshold, "Improved",
if(Z-score < -1 * threshold, "Declined",
"Stable")))

Where threshold refers to the cut-off z-score value for each descriptor. Commonly used thresholds are 1 or 1.5.

Note: in this example, all of the metric values are scored such that poorer options are scored lower and more positive options are scored higher. If your metrics are scored in the reverse direction, you will need to adjust your z-score calculation to account for this.

Outcome 2: Apply performance standards to z-scores

To assign performance standards to the z-score flags, navigate to the performance standards tool on the builder interface and create a new performance standard for the relevant fields.

Outcome 3: Visualize the z-scores using the dashboard builder

Use tiles to display an individuals' most recent fatigue flag

Tiles may be used to show a number of variables, including-

• The most recent score for an individual.
• The average score for the group.
• The number of individuals with wellness flags.

Rulesets may be applied to the tiles to highlight values of interest.

1. In the Dashboard builder, create a new dashboard or open an existing dashboard.
2. Ensure that the wellness event form is a data source in the dashboard.

   

3. Create a new tile using the Add Widget button.

   

   • In the Data section, select the appropriate data source and apply data filters if applicable (e.g. if there is a select box widget in the dashboard to filter for an individual).
   • In the Value section, select the flag field and select the appropriate data type and aggregation.

     

4. Navigate to the Rulesets menu and set up a ruleset for wellness flags. Ensure that the data type matches the data type selected in the previous step.

   

5. Return to the tile and in the Conditional Formatting section, enable conditional formatting and select the ruleset.
   

   

Use tables to display z-score flags for a team

Tables may be used to show flags for each person, or to highlight people who have flagged in particular areas by using filters.

Example: Using a table to display each athletes' most recent wellness scores and flags.

Example: Using tables with data filters to list athletes who are flagging for each wellness metric.

1. In the dashboard, create a new data source for the wellness event form which only includes data for today.

   

2. Create a table widget using the Add Widget button.

   

   • In the Data section, select the appropriate data source. Edit Data filters and add a filter to include records where the Z score is less than -1.

     

   • In the Columns section, add a column for the person’s name and for the score. The resulting dashboard will show only players who have a z-score of less than -1 today.

     

Outcome 4: Set up performance alerts based on z-scores

To create performance alerts based on the z-score/flag values, navigate to the performance alert management tool through the administration site or main interface and create a new performance alert. Select the appropriate Monitored Athletes or Monitored Groups and Notified Users.

Test your workflow

• Ensure that your role has permission to view the event form in the main interface.
• Enter some test data for a period of time into the wellness event form against an account.
• Check that all of the z-score values and flags are calculated correctly in reports.
• Ensure that the flags are correctly colored by the performance standards in reports.
• Load the dashboard and check that the data is displayed correctly. If there are athlete or date selectors in the dashboard, check that the values change when you adjust the selected options.
• Adjust the performance alert to monitor the test account and notify you. Enter a test record which would be considered a z-score flag. Check that you receive a performance alert and that it contains the information you need.