MEDICA DEEP DIVE: AI in diagnostics: a trustworthy companion
Menu

MEDICA DEEP DIVE: AI in diagnostics: a trustworthy companion

Image: A man is looking at a screen with imaging data; Copyright: Messe Düsseldorf

In this live talk on 21 September 2023 we focused on the topic of Artificial Intelligence (AI).

Especially in the field of imaging, AI technologies can be used more efficiently and precisely. This reduces the workload of radiologists and also leads to safer diagnosis.

However, it is essential that we understand AI systems and how they arrive at their recommendations. Especially in the field of medicine, this is important to increase user acceptance on the one hand and to ensure safety in treatment on the other.

In our MEDICA DEEP DIVE, we addressed AI solutions for imaging in practice and talked about how trust in AI technologies can be ensured.

Image: Smiling woman with long dark hair and a white blazer in a park - Dr. Narges Ahmidi; Copyright: Fraunhofer IKS

Dr. Narges Ahmidi is the Head of Department for Reasoned AI Decisions at the Fraunhofer Institute for Cognitive Systems IKS and summarizes trust as follows: "When I turn my back to this AI, (…) this thing doesn’t do something crazy when I leave it around."

Dr. Ahmidi compares trusting AI to trusting the autopilot on an airplane. Although the plane is being controlled by the autopilot for the majority of the flight, we fully trust it. This is due to the fact, that autopilot has been around for decades and has proven itself to be trustworthy. Explainability is a key word when talking about trust. "If you don’t understand it you don’t want to see it."

Furthermore, she also makes it very clear, that AI is not an isolated "magic machine" in a basement, but that working with AI always means being in a constant exchange with doctors and healthcare professionals. This is crucial as the AI "mimics the work of doctors" with real data from hospitals.

Image: Smiling man with short graying hair and beard in a gray pullover - Jan Beger; Copyright: private

Jan Beger is Senior Director Digital Ecosystem EMEA at GE Healthcare Information Technologies GmbH & Co. KG and knows that safety is the number one objective in the industry. While not wanting to rush into any new technology, Beger sees great potential in reducing medical errors with the support of AI and thus contributing to patients safety. Nonetheless, he very much agrees with Dr. Ahmidi on the issue of black box as he emphasizes the importance of explainability.

"Nobody would board a plane if there was no pilot on board."

Image: Physician with short black hair, glasses and a white coat sitting at a desk - Dr. Martin Bansmann; Copyright: beta-web

"There is not an AI doctor sitting here doing my job completely", assures us Dr. Martin Bansmann, Chief Physician at the Institute for Diagnostic and Interventional Radiology. Although AI is very useful for radiological image computing for example, AI is there to support human professionals and not replace them. Another benefit of AI are the possibilities to reduce "bullshit jobs", like documentation for example. This in return means more time for patients.

All of our speakers agree that one of the biggest challenges for the implementation of AI into various fields of healthcare is the availability of complete Data Sets to train algorithms. As Dr. Bansmann puts it, complete and usable data sets "are the holy grail right now".

MEDICA DEEP DIVE is now available as video-on-demand

You can watch our MEDICA DEEP DIVE "AI in diagnostics: a trustworthy companion" as video-on-demand here.

The video is available for users who are registered at the MEDICA website. If you are already registered, you can use the form below to log in.

Otherwise, please use this button to visit the registration page. Registration on the website is free.

Register here