Roundtable No°4 | 2024: From Data to Decisions - AI and CDSS in healthcare
Note: The next roundtable will take place on June 27 on the topic “Seamlessly connected: SDC and the future of interoperability in healthcare”. If you are interested, please register here.
After the 4th Patient Monitoring Roundtable 2024, the first roundtable half of 2024 is now behind us. What an exciting year! Once again, the evening, which was entitled “From Data to Decisions: How AI and CDSS are transforming healthcare” was dedicated to a much-discussed topic and was extremely interesting and informative.
As a great prelude to the event, Prof. Müller-Birn, Head of the Research Group “Human-Centered Computing” at the Free University of Berlin, explained the basics of clinical decision support systems and “Human-AI-Collaboration” in a keynote speech The keynote speech can be found here in full length.
The following points were identified as take-home messages:
A human perspective on AI needs to be integrated
Research on AI Integration is in its infancy. We need more case studies
Transparancy helps people to form better mental models about the sustainability of AI for the task at hand
Tansparency can be achieved by expainin a model’s decision or by describing the datasets characteristics used to train the model.
Preliminary results from one of the workshops
In the second part, we focused on the “instruction leaflet” that should be included for the use of AI and clinical decision support systems (CDSS) to inform clinical staff. In small groups, we used a so-called “model card” to discuss the comprehensibility and usefulness of various types of information and presentation.
It became apparent that, with regard to the properties of the machine learning models used, little transparent and well-prepared information is available for clinical decision support, analogous to a package insert for a drug. The information (e.g. with / on the basis of which population the model was trained, how accurate it is) should be presented in a user-centered way so that this information is also accessible in everyday clinical practice for quick decision-making and facilitates an assessment of the underlying model.
In summary, the evening showed that the use of artificial intelligence in the clinic offers numerous possibilities and can provide significant added value, especially in decision support. However, a number of conditions must be met. We would like to thank all participants for their contributions and perspectives on this extremely exciting roundtable! And above all, we would like to thank Prof. Müller-Birn for her excellent keynote speech!
The Patient Monitoring Roundtable is organized and conducted by INCH e.V. (Initiative for Innovation and Collaboration in Healthcare e.V.) in cooperation with the Institute for Medical Informatics at Charité - Universitätsmedizin Berlin.