file
Covance Blog - Sharing Innovation in Drug Development
    Publish
     
      • Using data visualization tools on SEND datasets: A graph is worth a thousand data points


        Posted by
        Published On May 20 2019, 4:11 PM

        SEND [Standard for Exchange of Nonclinical Data] is more than just a tool to facilitate nonclinical data submissions to the FDA. SEND datasets are rich in information, albeit in a form that’s time-consuming for non-experts to parse. With the right visualization tools, SEND data sets can inform nonclinical programs and yield important insights.  

        How Will Data Visualization Tools Help Me to Use SEND?

        Controlled terminology is useful in presenting data, giving investigators the ability to drill down into study details with a fine-tooth comb. Using a visualization tool can dramatically reduce the time needed to pour through the anatomic pathology results. Since these data are qualitative in nature, the ability to quantify, color-code, group and sort facilitates easier data viewing for pattern recognition, trend spotting and outlier identification.

        Here are some examples of how visualization can make reading data easier and quicker:

        Example 1: Visualization of Side Effects Reported in SEND Dataset

        Figure 1 below shows potentially concerning observations from a study: five male subjects suffering from ‘Nephropathy, chronic progressive — Severe’; the details of which were expanded into a table for the subjects affected.

        Figure 1. Visualization of observed side effects reported in SEND dataset.

        Example 2: Report of Side Effects in a Single Research Model

        Figure 2 shows a singular subject from the study that investigators selected to look at other tissues or findings further. This feature allows investigators to profile individual subject results and findings in order to identify trends in clinical side effects.

        Figure 2. Report of side effects observed in a single research model.

        Example 3: Summary of Clinical Parameters from SEND Dataset

        Before visualization

        Without any data visualization assistance, the SEND study below (Table 1) shows one clinical parameter (CP) per page and one sex per page. But this page is not the only page of data for males for this parameter; in fact, reporting all summary results for males and females for three parameters takes 11 pages.

        Table 1. Summary of clinical parameters from SEND dataset.

        Individual results for the same three parameters span 24 pages. All CP data in this study are reported on more than 500 pages of the CP contributor report. The entire study report was more than 1500 pages long. This length makes it very time consuming to follow the results for one subject of interest; and interpreting the data becomes non-intuitive.

        After visualization

        To make this data easier to grasp, it has been turned into a visualization. In Figure 3, the same three parameters from the above dataset are summarized into a single screen, sorted by both sexes combined (top) and separated (bottom).

        Since the visualization tool is interactive, scientists can easily shift between summary and individual data, individual research models, various parameters, etc. You can also select just one subject to highlight its data across all associated visualizations.

        Figure 3. Visualization of SEND dataset.

        Example 4: Clinical Observations from a SEND Dataset

        Before visualization

        In another sample data set, Table 2 indicates the number of subjects with a finding at any point during the study; however, it does not indicate how many times a finding was called.

        The summary table was split by shipment, leading to the findings in each category being separated by >10 pages. The summary table is 20 pages with individual tables lasting for 424 pages with the entire study report lasting more than 2700 pages.

        Table 2. Summary of clinical observations from SEND dataset.

        After visualization

        Using data visualization tools, all skin and pelage is represented in a single table (Figure 4). Bands indicate how many times the finding was made for each subject. All findings are now visible on one screen, and the user can filter the results by selecting other groups, parameters, or individual subjects.

         

        Figure 4. Visualization of data from a SEND dataset

        What Data Visualization Tools Can I Use?

        At Covance, we use data visualization tools internally to help us read and understand our own datasets. And you can do the same with your datasets.

        Scores of visualization tools are on the market that can accept raw data and turn lengthy study reports into intuitive, interactive graphics. You’ll want to do your own research and remember to look for a tool that fits well within your existing data systems.

        If you have more questions about SEND or data visualization, please connect with us. Or learn more about SEND in our SEND Education Center.

      This entry was posted in Drug Development and tagged , , by Bookmark the permalink