DESIGNING AN INTERACTIVE DESKTOP DASHBOARD

AI Quick Summary

Background The metadata repository we developed in the context of the Screen4Care project is to be used by clinicians to accelerate rare disease diagnoses using Machine Learning. While most of the infrastructure is in place, we are looking for creative solutions for designing a UI/UX interface. In this scenario, the clinicians need to understand which patient data fields – hereafter referred to as metadata - is compatible and in the same format between hospitals. The metadata repository never stores actual patient data, but only the information which patient data fields and format are available (see attached sample data). We are seeking an innovative web-based user interface on a desktop that enables clinicians to identify such metadata effectively. The ultimate goal is for clinicians to pinpoint which metadata and hospital sources are available, so they can use them in machine learning studies. For an overview, watch this VIDEO.   Disclaimer: This video is intended for informational purposes only. While some graphics are included to support understanding, they are not meant as design references or visual guidelines. Furthermore, the video discusses features, but they are in this context the same as metadata. Music: https://www.bensound.com. Artist: Benjamin Tissot. License code: 6CSP0XDCP2UK7SDR.   User Needs & Use Cases Clinicians will need to have easy access to the following interface functionalities: 1. View Hospital Summaries View basic hospital details: name, location, number of patients, quantity of patient records and metadata available 2. Compare Two Hospitals for Metadata Compatibility Search hospitals by name or location Select two hospitals and see a visual summary of their metadata compatibility For example, when selecting “Charité – Universitätsmedizin” and “University Medicine Göttingen”, the UI/UX interface not only shows the common metadata, but also indicates which metadata are considered compatible. They both share the metadata Date of Birth and the format “Full date”. 3. Visualize Partial Compatibility Show where the metadata is partially compatible and provide explanations Similar example to 2nd use case. If “Charité – Universitätsmedizin” and “Odense University Hospital” are selected, there are significant differences between the metadata and format. For example, Date of Birth are partially compatible since the format “Full date” can also be represented as “Year” at the cost of losing additional information. In some cases, hospitals use clinical vocabularies for standardized documentation of laboratory test results, diagnoses, and many more, such as SNOMED-CT, LOINC, and ORPHANET. However, they are not fully compatible with each other. Therefore, it is sufficient to declare they are partially compatible. 4. Compare More than Two Hospitals Enable multi-hospital comparison without overwhelming users This use case is an extension of 2 and 3, where the user can include additional hospitals for in depth comparison. On average, we expect the user to interact and compare with around five hospitals. If the design is too complex and overwhelming for the users, it should be possible to swap to a simplified design. 5. Smart Suggestions Recommend additional hospitals with overlapping compatible features Once the user selects at least one hospital, the metadata repository can recommend additional hospitals for the user to include. The recommendations should always be available, but at the same time do not distract the user. 6. Search by Patient data fields Enter specific metadata (e.g., date of birth, BMI) and see which hospitals have them All hospitals display available metadata allowing the user to consider which metadata must be used for their Machine Learning study. For example, older patients may be susceptible to show certain rare diseases, therefore we only want hospitals that have the metadata “date of birth”.   7. Grouped Hospital View Group hospitals by the amount of compatible metadata, sorted descending Instead of showing hospitals individually, there should be an option to display grouped hospitals by the number of compatible metadata. For example, “Charité – Universitätsmedizin” and “University Medicine Göttingen” could be grouped since they share many compatible metadata.   What is not in the scope? While some use cases allow interpretation and creative input, we outline the following assumptions for clarity: There are no technical limitations for this task. Your primary focus is to design the interface to allow the clinicians to identify compatible features between hospitals to facilitate machine learning studies. Machine learning studies are outside the scope of this project. The idea is that the clinicians require the following information: Which metadata is available and compatible given the selected hospitals? If I remove some hospitals, how much metadata will I potentially gain? It is essential for the clinicians to interact and explore with your interface to understand this information before performing a machine learning study.   Criteria and incentives The selection team will be composed by Prof. Richard Roettger, University of Southern Denmark, PD Dr. med. Jana Zschüntzsch, Universitätsmedizin Göttingen, Rudolf Mayer, SBA Research Vienna, and Nicolai Dinh Khang Truong, PhD Student, University of Southern Denmark. All submitted ideas will be reviewed by the selection team. They will select the design or design elements that optimally align and support best the goal, not necessarily a single design or a specific number of them. Participants whose design elements are chosen for implementation will receive specific acknowledgment on the meta-data repository webpage. The selection team will gladly provide a letter of recommendation to requesting participants. Co-authorship in a scientific article may be offered to participants whose elements have been selected and make up a substantial portion of the visualization interface, if information about the meta-data repository is deemed publishable.   How to submit your ideas? Your submission can entail "mock-ups", either computer graphics or hand-drawn, as i.e. static images of what it could look like, coupled with explanations if there is interaction or dynamic behavior envisioned. For computer graphics, you can use your favorite interface design software (Figma etc) and submit your solutions in jpg, png or pdf. For hand-drawn sketches, ensure that the writing and drawing are understandable. It should also be clear how the user can interact with your interface. We expect the mock-up to be optimized for display on a standard 1920 x 1080 monitor. It is not necessary to create mock-ups for tablets or smartphones. Just click on the button “Submit” (top right) and get started. Contact nitru@imada.sdu.dk

Challenge Metadata

• Challenge Name: DESIGNING AN INTERACTIVE DESKTOP DASHBOARD
• Date: 5/12/2025 11:57:36 AM
• Administrators:
  • Sem Mattli  | 
  • Adrian Bachofen  | 
  • Thu Nguyen  | 
  • nadine c martin

Challenge Description

Background

The metadata repository we developed in the context of the Screen4Care project is to be used by clinicians to accelerate rare disease diagnoses using Machine Learning. While most of the infrastructure is in place, we are looking for creative solutions for designing a UI/UX interface.

In this scenario, the clinicians need to understand which patient data fields – hereafter referred to as metadata - is compatible and in the same format between hospitals. The metadata repository never stores actual patient data, but only the information which patient data fields and format are available (see attached sample data). We are seeking an innovative web-based user interface on a desktop that enables clinicians to identify such metadata effectively. The ultimate goal is for clinicians to pinpoint which metadata and hospital sources are available, so they can use them in machine learning studies.

For an overview, watch this VIDEO.  

Disclaimer: This video is intended for informational purposes only. While some graphics are included to support understanding, they are not meant as design references or visual guidelines. Furthermore, the video discusses features, but they are in this context the same as metadata.

Music: https://www.bensound.com. Artist: Benjamin Tissot. License code: 6CSP0XDCP2UK7SDR.

User Needs & Use Cases

Clinicians will need to have easy access to the following interface functionalities:

1. View Hospital Summaries

• • View basic hospital details: name, location, number of patients, quantity of patient records and metadata available

2. Compare Two Hospitals for Metadata Compatibility

• • Search hospitals by name or location
  • Select two hospitals and see a visual summary of their metadata compatibility
    • For example, when selecting “Charité – Universitätsmedizin” and “University Medicine Göttingen”, the UI/UX interface not only shows the common metadata, but also indicates which metadata are considered compatible. They both share the metadata Date of Birth and the format “Full date”.

3. Visualize Partial Compatibility

• • Show where the metadata is partially compatible and provide explanations
    • Similar example to 2nd use case. If “Charité – Universitätsmedizin” and “Odense University Hospital” are selected, there are significant differences between the metadata and format. For example, Date of Birth are partially compatible since the format “Full date” can also be represented as “Year” at the cost of losing additional information. In some cases, hospitals use clinical vocabularies for standardized documentation of laboratory test results, diagnoses, and many more, such as SNOMED-CT, LOINC, and ORPHANET. However, they are not fully compatible with each other. Therefore, it is sufficient to declare they are partially compatible.

4. Compare More than Two Hospitals

• • Enable multi-hospital comparison without overwhelming users
    • This use case is an extension of 2 and 3, where the user can include additional hospitals for in depth comparison. On average, we expect the user to interact and compare with around five hospitals. If the design is too complex and overwhelming for the users, it should be possible to swap to a simplified design.

5. Smart Suggestions

• • Recommend additional hospitals with overlapping compatible features
    • Once the user selects at least one hospital, the metadata repository can recommend additional hospitals for the user to include. The recommendations should always be available, but at the same time do not distract the user.

6. Search by Patient data fields

• • Enter specific metadata (e.g., date of birth, BMI) and see which hospitals have them
    • All hospitals display available metadata allowing the user to consider which metadata must be used for their Machine Learning study. For example, older patients may be susceptible to show certain rare diseases, therefore we only want hospitals that have the metadata “date of birth”.  

7. Grouped Hospital View

• • Group hospitals by the amount of compatible metadata, sorted descending
    • Instead of showing hospitals individually, there should be an option to display grouped hospitals by the number of compatible metadata. For example, “Charité – Universitätsmedizin” and “University Medicine Göttingen” could be grouped since they share many compatible metadata.

What is not in the scope?

While some use cases allow interpretation and creative input, we outline the following assumptions for clarity:

• There are no technical limitations for this task. Your primary focus is to design the interface to allow the clinicians to identify compatible features between hospitals to facilitate machine learning studies.
• Machine learning studies are outside the scope of this project. The idea is that the clinicians require the following information: Which metadata is available and compatible given the selected hospitals? If I remove some hospitals, how much metadata will I potentially gain? It is essential for the clinicians to interact and explore with your interface to understand this information before performing a machine learning study.

Criteria and incentives

The selection team will be composed by Prof. Richard Roettger, University of Southern Denmark, PD Dr. med. Jana Zschüntzsch, Universitätsmedizin Göttingen, Rudolf Mayer, SBA Research Vienna, and Nicolai Dinh Khang Truong, PhD Student, University of Southern Denmark. All submitted ideas will be reviewed by the selection team. They will select the design or design elements that optimally align and support best the goal, not necessarily a single design or a specific number of them.

Participants whose design elements are chosen for implementation will receive specific acknowledgment on the meta-data repository webpage. The selection team will gladly provide a letter of recommendation to requesting participants. Co-authorship in a scientific article may be offered to participants whose elements have been selected and make up a substantial portion of the visualization interface, if information about the meta-data repository is deemed publishable.

How to submit your ideas?

Your submission can entail "mock-ups", either computer graphics or hand-drawn, as i.e. static images of what it could look like, coupled with explanations if there is interaction or dynamic behavior envisioned.

For computer graphics, you can use your favorite interface design software (Figma etc) and submit your solutions in jpg, png or pdf. For hand-drawn sketches, ensure that the writing and drawing are understandable. It should also be clear how the user can interact with your interface.

We expect the mock-up to be optimized for display on a standard 1920 x 1080 monitor. It is not necessary to create mock-ups for tablets or smartphones.

Just click on the button “Submit” (top right) and get started.

Contact

nitru@imada.sdu.dk