Physicians have the job to diagnose and, subsequently, treat medical conditions. For that they rely on tools and their knowledge to perform diagnoses. The readout of tools allows insights into biological processes. The integration of those readouts reflects an overall probability for a certain medical condition being present. Since machine learning algorithms are able to perform such an integration (often with higher reproducability than physicians), it is reasonable to assume thatmachine learning can be used for supporting physicians in making decisions about the diagnosis of medical conditions. Never the less, machine learning has not (yet?) been able to gain a foothold in daily clinical practice.
In this talk we will explore some of the reasons why this hasn’t been the case and discuss what needs to be changed in the application of machine learning algorithms so that this might become more feasable in future healthcare.
