Developing an AI-based diagnostics application

• Supporting medical diagnoses through AI is very complex, as it combines high demands on transparency and compliance. 

• Zühlke developed a solution in an interdisciplinary team - using proven processes from the regulated development of software. 

• The solution supports doctors in making difficult decisions and enables them to spend more time on patients themselves.

When AI is used for diagnosis – it has to be medical compliant and transparent making software development very complex. Zühlke delivered the solution in six months.

Demanding interdisciplinary task

An international pharma company wanted to provide physicians with a diagnostics application based on Artificial Intelligence (AI) that supports the differential diagnosis of specific diseases. However, as the recommendations of this software are affecting the treatment of patients, it has to be implemented in a regulatory compliant way. With AI – this is a demanding interdisciplinary task.

From prototype to market ready application

The Zühlke team consisting of experts for machine learning, data engineering, software development as well as regulatory experts performed the operationalization of the existing AI research prototype. In six months, Zühlke transformed the AI prototype – a model based on machine learning – into an application that is market ready and regulatory compliant.

Safe, Sustainable and Regulatory Compliant

In order to create the diagnostics software safe, sustainable and in a regulatory compliant way, Zühlke successfully applied its own process for the development of AI-based medical solutions. The result is an application that provides physicians not only with a diagnosis of the disease but can also explain, which factors were important for this prediction which creates trust and transparency.