Data Challenge - Driving Evidence-Based Antibiotic Prescribing: Machine Learning-Enabled Point of Care Antibiograms for Improved Patient Care

Aim: To use machine learning and data from the ATLAS project to design a point of care antibiogram application (job aid) for medical prescribers. Additionally, the project aims to compare the burden of antimicrobial resistance (AMR) between low middle and high income countries using Kenya, South Africa and the United Kingdom as case studies. The global burden of antimicrobial resistance AMR has earned a place on the World’s list of high priority health problems. In 2019, AMR was responsible for approximately 5 million deaths worldwide (CDC 2021). Central to the fight against AMR is antimicrobial stewardship through the use of antibiograms. Easy access to antibiograms is critical in making real time decisions on empirical antibiotic therapy (EAT). Smartphone applications have been shown to ease access to information, including knowledge of prescribing antibiotics (Fralick M et al). The value of artificial intelligence The idea of artificial intelligence (AI) has completely changed how many 21st-century industries work. Computers may learn without being explicitly programmed thanks to the field of machine learning, a branch of artificial intelligence. This suggests that as an application is used more often, it will be able to forecast outcomes with more accuracy. This technology can therefore help prescribers provide suitable EAT with the proper data. Antimicrobial Testing Leadership and Surveillance program The Antimicrobial Testing Leadership and Surveillance (ATLAS) program provides data on trends in antimicrobial resistance in several countries including both high and low income countries. The program uses Clinical and Laboratory Standards Institute (CLSI) breakpoints and has worked on over 790,000 isolates. The ATLAS database is updated every 9-12 months making it an appropriate source for development of antibiograms as a means for increasing access and usability of the surveillance data.