Clinical evaluation of AI-driven decision support for the treatment of infections caused by antibiotic-resistant bacteria

Infections from antibiotic-resistant bacteria threaten human health and are associated with 5 million deaths, yearly. Accurate diagnostics are vital to ensure timely treatment with efficient antibiotics.

However, the urgency of life-threatening infections often requires immediate action before diagnostics results are available (‘empirical treatment’).

This is prone to failure and leads to unnecessary antibiotic consumption.We have developed a decision support system (DSS) that, based on patient information and available diagnostics results, guides the choice of antibiotic treatment.

The DSS is based on artificial intelligence (AI) and trained on empirical data from millions of previous patients with bacterial infections.

Preliminary evaluation shows that the DSS improves upon empirical treatment and has the potential to reduce mortality and unnecessary antibiotic consumption.In this proof-of-concept project, we will implement and evaluate the DSS in clinical settings using three major hospital laboratories in Sweden as test beds.

We will estimate the accuracy of the DSS when used close to the clinical routine and ensure that the provided guidance is robust and fair.

We will, finally, estimate the health and social benefits of using AI-guided decision support for the treatment of bacterial infections.This project will pave the way for novel evidence-based diagnostics methods that utilize AI to meet the growing burden of infections from antibiotic-resistant bacteria.