AI cannot prevent diagnostic errors
About one in ten diagnoses is incorrect. Even AI cannot prevent this, as the world's first study on AI-based diagnostic systems in acute medicine shows. The study was published as a part of the NRP 77.
Misdiagnoses are one of the most common and costly medical problems worldwide. Many hospitals are therefore turning to AI and making diagnoses with the help of Computerised Diagnostic Decision Support Systems (CDDSS for short).
However, the merits of CDDSS are controversial. In the first study worldwide, Prof Dr med Wolf Hautz from Inselspital Bern examined the quality of these systems in acute medicine as part of the National Research Programme NRP 77. As head of emergency medicine, he and his team frequently face misdiagnoses. In emergency departments, they are faced with treating many different patients with a variety of symptoms under great time pressure.
Surprising results of the world's first AI-diagnosis study
Following a study involving over 1,000 patients, Prof Dr med Wolf Hautz concludes:
AI-based diagnostic support has no measurable effect on patients in emergency medicine. Regardless of whether you look at medical, economic or procedural differences.
Prof. Dr. med. Wolf Hautz‘Research is still in its infancy’
For AI to realise its potential, AI systems and diagnostic processes need to be developed further. In the joint press release from the Insel Gruppe and the University of Bern, Hautz therefore calls for new approaches:
We need to pursue other approaches to improve the quality of diagnosis - and in particular significantly intensify research on this topic, which is currently in its infancy.
Prof. Dr. med. Wolf HautzThe Swiss National Science Foundation (SNSF) is funding the establishment of a working group on ‘Collaborative Decision Making’ at the Inselspital's Department of Emergency Medicine.
The study on AI-based diagnostic systems in acute medicine was published in January 2025 in the journal ‘The Lancet Digital Health’. The study was conducted as part of the National Research Programme NRP 77, which focuses on digital transformation.