Overview
The correlate panel is a powerful tool designed to enhance the analysis of clinical trial data by providing intuitive and informative scatter plots. Users can explore the relationship between two variables, and gain valuable insights from the data. By visualising correlations between key parameters, you gain a deeper understanding of treatment response, safety profiles, pharmacokinetics, and baseline characteristics.
Use Cases
Here are some examples of questions you can answer using Correlations:
Medical Monitoring:
Is there a correlation between the dosage of a medication and the efficacy outcomes observed in the study population?
Are there any relationships between concomitant medications levels and treatment outcomes?
Do specific subgroups or outliers within the study population exhibit unique treatment responses or outcomes?
Safety Monitoring:
Are there any correlations between laboratory parameters and the occurrence of adverse events?
How does the risk of drug induced liver injury correlate with different treatment arms?
Does the occurrence of specific adverse events differ based on patient demographics or baseline characteristics?
Clinical Operations
Are there any correlations between site performance metrics like patient screening and dropout rates, and study outcomes?
Are there any correlations between patient demographics and study compliance or outcomes?
These are just a few examples, as the possibilities for data exploration using Correlations are extensive. This tool empowers users to create customised visuals, facilitating deeper insights and data-driven decision-making in clinical trial monitoring and analytics.
Quick Start
With just a few clicks, you can easily create a Correlation. Let's build a simple Correlation together, following the eDish approach, imagine you want to answer the following question:
Is there any correlation between ALT and Total Bilirubin levels for the patients and does it vary depending on the treatment group?
Step 1: Choose Events
Start by selecting two relevant events: one for the x-axis and another for the y-axis. In this case we want to visualise the maximum ALT and Total Bilirubin values for each patients.
Initially the different events are combined into a single data point using the Patient ID of the patients. We call this the "join key". You can modify, add or delete these join keys as required. For more information, refer to the section on join keys in "Advanced Features".
Step 2: Choose a Cohort for your Correlation
Cohorts define a subset of patients. For our example, let's select patients who have been exposed at least once to the investigational drug.
Step 3: Choose Filters
Filters exclude unwanted data and can select specific subsets of data to highlight trends or anomalies. Let's select the ALT and Total Bilirubin levels taken during the 30 day follow-up visit, for this we apply the following filter:
Step 4: Choose Segments
Segments allow you to group your data points into different categories. In the scatter plot, segments are represented using different colours. Let's divide our dataset using the treatment arm.
Basic features
Statistics
Hover on the regression line to see additional information about your dataset. You can explore three key statistics to further analyse your data: the linear regression line, the p-value (where the t-value is computed by dividing the regression line slope by the standard error), and the R-squared value.
Advanced features
Configurations
Access advanced customization options by clicking the Settings button.
Adjust settings for scaling the x and y axes, reference lines, and regressions and optimize combinations for clarity and ease of result analysis.
Continuing on with our example from the "Quick Start" section, we obtain the following eDish plot by moving the reference lines and rescaling the x and y axis.
Join keys
Choose the join key to define how multiple events are combined into a single data point. Events are, by default, aggregated using the Patient ID. Alternatively, you can select a different property, such as the treatment arm, to create separate data points for each group.
