The human brain can grasp the meaning of an image in as little as 13 milliseconds, and people learn more deeply from words and pictures than from words alone. Without effective visualisation, data can be overwhelming and difficult to interpret. However, with the right design, visualisations can transform complex data into clear, accessible, and actionable insights.
Why data visualisation matters
The human brain can grasp the meaning of an image in as little as 13 milliseconds,1 and people learn more deeply from words and pictures than from words alone.2 Without effective visualisation, data can be overwhelming and difficult to interpret. However, with the right design, visualisations can transform complex data into clear, accessible, and actionable insights.
Bringing complex data to life
At AstraZeneca, we believe that data visualisation is key to revealing insights and effectively communicating complex information.
Our data visualisation tools don’t just make clinical trial results easier to interpret, they help unlock deeper insights into patient outcomes. From single trials to multi-study landscapes, we are developing innovative visuals that turn abstract data into accessible, actionable understanding.
Layered patient outcomes at a glance
The Maraca plot1,2,3,4 is designed to visualise hierarchical composite endpoints (HCEs) with greater clarity and precision. The HCE combines different types of patient outcomes, from major clinical events like death and hospitalisation to gradual changes such as kidney function decline, and the Maraca plot shows the complete study results in a single intuitive visual.
Mapping treatment effects across trials
While the Maraca plot focuses on a single trial, the Sunset plot offers a broader view. It visually explores various ‘what-if’ scenarios, such as identifying combinations of hazard ratio and mean difference in continuous distributions that result in the same win odds (similarly coloured regions). The overall treatment effect on the HCE, resulting in the same win odds, could be achieved by different combinations of treatment effects on the time-to-event outcomes of the HCE (estimated by a hazard ratio) and the treatment effect on the continuous component (mean difference). Lighter regions show the higher values of win odds (i.e. larger treatment effects) while the darker areas are for lower win odds values. The shaded area shows possible combinations of treatment effects for kidney HCE which is a particular case of an HCE where all outcomes are highly correlated (treatment effect consistency is expected on all outcomes). Therefore, the plot enables us to represent the universe of all possible clinical trials with kidney HCE as the primary endpoint.
This ‘big picture’ approach helps researchers understand both the results of different studies and how large the treatment effects are.
Tracing patient journeys over time
To simplify safety readout in large clinical studies, the Tendril plot7 summarises and explores the adverse event (AE) data, capturing not just the frequency of AEs, but also their timing and distribution between treatment arms. In a single visualisation, this plot makes it possible to graphically show important treatment differences with preserved temporal information, across an entire clinical trial.
Contribution of components
The Dustin plot, a combined bar and forest plot, consists of two panels; a bar plot of wins, losses and ties as additional components are added cumulatively using the severity order from top to bottom, and the forest plot with the win odds and win ratio corresponding to the same cumulative sequence.
Usually, a single number cannot communicate all the intricacies of the treatment effect; this visualisation allows for a better representation of the treatment effect and therefore a more nuanced way of explaining it to patients, prescribers, and other stakeholders.
Visualising group differences
The 2D mosaic plot is a simple visual tool to compare and understand how different groups are distributed across categories such as treatment arms and outcomes. For example, areas where the active treatment group has a higher proportion of favourable outcomes than the control group will stand out, helping researchers visually assess where the active treatment may be outperforming the control.
The real-world impact of data visualisation
The simplicity of the maraca plot makes it a powerful tool for interpreting data from e.g. chronic kidney disease (CKD) clinical trials.
CKD is a progressive, irreversible condition that places a significant burden on patients and healthcare systems.8 Clinical trials in CKD often use HCEs to capture the full clinical picture, prioritising the most severe events such as death or kidney failure, while also considering less severe but meaningful changes such as decline in eGFR rate. The maraca plot displays all these components in a single, integrated view, showing both the order of clinical importance and the size of each component’s contribution. By structuring clinical outcomes in this way, the plot allows researchers to see where treatments are making a difference across a spectrum of outcomes and support faster, more informed decision making.
Novel endpoints for CKD clinical trials require novel visualisation methods since traditional methods of presenting data often fail to capture the full impact of treatment, making it challenging to interpret the treatment effect. The beauty of the maraca plot is that it was specifically designed to visualise hierarchical composite endpoints and therefore can reveal small differences in the drivers of the treatment effects in a single, compelling visual.
Working to shape the future of data-driven healthcare
At AstraZeneca we are exploring the use of visualisation and how it can shape future directions:
- We are increasing the use of visualisation as a better way of presenting the data and consider it as an essential part of the statistical training for our researchers.
We are driven by a fundamental belief: the visualisation of data is key to revealing insights and effectively communicating complex information. Guided by this principle, we want to unlock the full potential of our data and drive innovation in how we approach scientific challenges. We are fostering a data-literate culture where employees can confidently upskill, create, interpret, and leverage impactful visualisations to extract potentially meaningful insights. The ambition spans simple graphical explorations, interactive dashboards to more complex novel charts and animations.
- We innovate to create novel ways of expressing data in collaboration with our external partners and share our work freely for everyone’s use because of our commitment to collaborative science.
In addition to increasing the use of existing visualisations, we are constantly exploring novel ways of expressing the data and creating new visualisation approaches. Our researchers have been the forefront of developing several new types of visualisations, like maraca, tendril, and Dustin plots. This spirit of innovation is further amplified by our commitment to open sharing and collaboration, both within AstraZeneca and with the broader scientific community. By developing open-source R packages and engaging in knowledge exchange, we're contributing to and benefiting from collective advancements in the field.
- Generative AI and the future of visualisations
With a well-articulated data visualisation strategy, we are not just changing how we view data at AstraZeneca – we're shifting how we think about and solve complex scientific problems. Our goal is to create a truly data-driven culture where visualisation isn't just a rarely used tool, but a fundamental aspect of our scientific approach, leading to faster discoveries, more informed decisions, and better outcomes for patients worldwide. Looking ahead, we expect data visualisation to become even more intuitive and conversational as technologies such as generative AI mature. We see visualisation as more than a tool for displaying results. It is a critical part of how complex scientific thinking helps drive faster insights and better decisions across trials and research programmes.
