TABLE OF CONTENTS

• How to extract data for group level data 
• How to extract data for the comparison
• Automatic data extraction from table with AI model

How to extract data for group level data 

To extract group level data, must first extract all study groups/arms. For example, if you want to extract data by single study arm, extract all arms in the ‘Treatments’ tab. Next, go to the Outcomes tab.

In the Outcomes tab you will see a separate section entitled “per Treatments”. Below this, the tool recognized and imported all study arms with the fields to be extracted per single study arm. All you need to do is to extract data separately for each arm. 

Note that all the data above the ‘per Treatments’ section are common for all of the study arms, e.g. Name, definition or unit, and you have to extract it once.

How to extract data for the comparison

Once you have extracted data per individual study arm, you can also add data for comparison. To do this, go to the 'Comparisons' section below the arm-level data and click the 'Add' button. You will then see a list of possible comparisons that have been generated based on previously generated study arms. Select the comparison for which you need to extract data. 

Once you have selected a comparison pair, you will see a list of fields for each comparison. Extract any relevant data. 

The last (optional) step is to create a connection between this comparison and predefined variable types such as subgroup or population types.

After you finish extracting data for a single comparison, you can select the next comparison to extract data for - click the 'Add' button in the blue 'Comparison' section and select the next pair of interventions.

Automatic data extraction from table with AI model

1. Click on the table from which you want to extract data and click on blue ‘Extract’ button.

In this example, our task is to extract outcome  data, including the name of the outcome (i.e., ACR70), the number of patients in cohorts, the number of patients with the event, and the percentage of patients with the event. 

2. Select from the list the type of section (outcome) you want to extract.  In our example, it will be ‘outcomes’.

3. Choose rows from tables which you want to extract. In our example it will be specific outcomes you want to extract.

You can use the 'Rotate the table' button if necessary.

4. Define the type of data that belongs to each column. You will have to select from the list of fields in your data extraction form.

 In our example, the first field includes the outcome names.The values that have been mapped and extracted from table are highlighted in yellow (see Extraction output in right side of the screen).

To one column, you can assign more than one data extraction field. This option will be useful in cases where one cell in the table includes two types of values, and you want to extract them separately to different fields in your extraction form. 

In our example, the second column includes two types of values: number of patients with the event and % of patients with the event. Select both field names from the list. 

Tip: The order of selected field names represents the order of values in the cell.

5. You can extract additional data for all outcome objects  by adding them manually in Extraction output. Select the relevant data extraction field from the list and specify what should be extracted (add text/value).In our example in the data extraction form, we have to extract the number of patients in each cohort. The same value should be extracted for all outcomes: Enter value in extraction output and  tool will automatically add this term to all 'outcomes objects'. You can see how the model extracted these fields in the next screen. Those manually added values are here highlighted in blue. 

You can see how the model extracted data from table in the next screen. Please accept values extracted automatically by model one by one.  You can also use ‘batch accept’ button.

RELATED ARTICLES  

1. Data extraction - introduction  
2. Data Extraction form - set up  
3.  Performing data extraction - start here to understand the process
4. How to extract data in a text field?  
5. How to extract data in a vocabulary field?