Sept. 20th-22nd, 2019, CAREUM, Zurich, Switzerland

Challenge B

Medical Needs Solvable By AI


The extent of digitization in today’s medicine is growing rapidly and so are the approaches to use these data for creating predictive models based on artificial intelligence (AI). Machine learning (ML) technology already has an impact on today’s medical practice. From image classification in radiographs, over epidemic outbreak prediction to genome sequencing, computer algorithms become more and more prominent in modern medicine. Projects of major companies like IBM Watson or Googles Deep Mind Health as well as numerous smaller privately and publicly funded research projects are pushing forward to close the gaps between medicine, mathematics and computer science.


However, not all approaches in this direction are success stories, like the failure of Googles Flu Trend (GFT) showed. We see several reasons for such failures. One is the lack of time dependence, were -like in the example of GFT- the models did not take into account the change of human (search) behavior over time. As a second cause we see that there is a tendency that such projects are mostly realized by IT experts and hence are more driven by the technical possible than by the medical meaningful. Often the already existing and well established knowledge in the domain field is integrated insufficiently if at all. For example many models are based on the available data, not on the data most suitable for the task. Finally to mention what some refer to as the big data hubris.  AI approaches and ML algorithms tend to perform at their best if trained on and used for tasks in which humans are already good in, rather than on problems were even the best human experts are struggling  to find suitable solutions. One of the main underlying reason for these problems seems to be a lack of combined access to medical expertise, clinical data and software development knowledge.

Actual Challenge

Start an open and transparent platform were medical experts, patients and IT developers can share their medical expertise, clinical data and software development knowledge to address medical needs solvable by AI (MNSBAI). The platform should allow:

  • Patients to
    • enter and manage their clinical data anonymously or in compliance with data privacy policies and in a somehow formalized and standardized way
    • donate some of their clinical data to a specific MNSBAI or
    • entirely to the public
  • Medical experts to
    • define MNSBAI that are medical meaningful, by defining the type of the task, define datasets and provide additional info on different features such as characteristics, time dependency, shortcomings, anomalies and specify quality characteristics….
    • enter and/or review ground truth (e.g. diagnoses) and data quality information (e.g. levels of confidence or credibility) to existing patients
    • “create” typical example-patients for specific MNSBAI
  • IT developers to
    • Get access to MNSBAI specific data sets
    • commit to make their solutions publicly available

Um unsere Webseite für Sie optimal zu gestalten und fortlaufend verbessern zu können, verwenden wir Cookies. Durch die weitere Nutzung der Webseite stimmen Sie der Verwendung von Cookies zu. Weitere Informationen zu Cookies erhalten Sie in unserer Datenschutzerklärung

The cookie settings on this website are set to "allow cookies" to give you the best browsing experience possible. If you continue to use this website without changing your cookie settings or you click "Accept" below then you are consenting to this.