Clin Res Cardiol (2022). https://doi.org/10.1007/s00392-022-02002-5

MedEx - platform for analysis and interactive visualization of HF patient data
M. Smieszek1, A. Kindermann1, H. Wilhelmi1, C. Dieterich1
1Klinik für Innere Med. III, Kardiologie, Angiologie u. Pneumologie, Universitätsklinikum Heidelberg, Heidelberg;

Motivation and Objective: In cardiology, we experience a qualitative and quantitative increase in clinical data capture in practice. The entire process is driven by the desire to gain deeper insight into a specific condition, to detect pathological conditions early on and to develop better and individual treatment. Any hypothesis generation and knowledge extraction require efficient, fast responding graphical solutions. Visualization can support clinicians in pattern discovery by enabling intuitive data interpretation, outlier detection, classification and regression features. Our goal was to create a lightweight tool to explore large amounts of bio-medical data. Importantly, this tool requires minimal resources and could be deployed in a clinical setting.

Approach: The Medical Data Explorer (MedEx) combines a web-based dashboard with a database backend. It is easy to deploy via Docker containers and enables exploratory data analysis by individual users. MedEx does not require knowledge of scripting or programming languages and is intuitive to use, making it suitable for clinician scientists and other personnel. The tool is available via Docker at https://hub.docker.com/r/aljoschak1/medex and GitHub repository at https://github.com/dieterich-lab/medex/releases/tag/v0.1.3.

Results: The software includes various types of visualization: scatter plot, bar chart, box plot, histogram, heatmap (figure 1), as well as the calculation of basic statistics: mean, median, standard deviation, standard error. Furthermore, it is possible to filter the underlying data by time of measurement, categorical or numerical parameters. Further external analyses could be conducted on selected and filtered data by the implemented download functionality. MedEx has been tested using >200,000 patient records from the UK Biobank database (number of numerical entities: 232, categorical entities: 203). We present selected examples based on UK Biobank data in Figure 1 and 2. A tutorial video can be found here: https://github.com/dieterich-lab/medex/blob/master/static/images/medex.mp4).

Conclusion: In summary, this tool allows medical professionals to navigate through medical data spaces and do their own exploratory data analysis. Relevant conclusions based on observed data patterns may emerge from exploration and drive future clinical research. 

C:\Users\roll\Desktop\Abstract_Grafiken\2021-11-24_08h09_40.jpg




C:\Users\roll\Desktop\Abstract_Grafiken\2021-11-24_08h11_02.jpg

https://dgk.org/kongress_programme/jt2022/aP1149.html