JMIR Publications

ABSTRACT

Background:

Different data visualization tools have been compared in previous studies, but very few focused on health-related research and operations in a case study using real-world data.

Objective:

Our study aims at comparing several visualization tools that are commonly used in health research by applying these tools to visualize two health datasets with different technical aspects. The goal of this study is to provide general suggestions for future visualization efforts in health-related scientific discoveries as well as biomarker operational management.

Methods:

Through a systematic search for the period between 2012-2022, we identified four commonly used visualization tools, including Tableau, Spotfire, PowerBI and R Shiny. We then compared these tools based on different technical aspects such as visualization outcomes (table, scatterplot, correlation plot, treemap, heatmap and geographic map), usability, Cloud compatibility, analytics integration, and other factors.

Results:

All four tools can generate very comparable visualization outcomes given standardized settings. Some tools may provide additional features than others. Regarding usability, all tools except R Shiny are considered easy to use based on eight parameters evaluated. Due to its scripting nature, R Shiny is not recommended for users with limited coding experience in developing a visualization platform. However, R Shiny can leverage many cutting-edge technical packages provided by the academic communities and offer great flexibility and customization. The tools evaluated all support connections to common Cloud-based data sources such as PostgreSQL, gSheet and Athena. Most tools support these connections without additional configurations except R Shiny. All four tools support analytics integration with R and Python, which allow for advanced analytics and machine-learning to be used. However, PowerBI requires some configuration for use of these scripting languages. These four tools are all capable of pre-processing data such as data transformation, null detection and outlier detection. As to performance, Tableau was faster than Spotfire and PowerBI to load the dataset and open workbooks, but it was the slowest for publishing a workbook. R Shiny was slower in creating plots because it needed to reload the dataset each time the tool was run.

Conclusions:

The four data visualization tools compared based on the various technical aspects we selected have their own unique advantages and limitations within the context of health-related research and operational management. While some tools are easier to use for certain features, most tools can create similar or comparable visualization outcomes. Users should match their use case against the capabilities of the tools.