Antimicrobial resistance surveillance through innovative ML/AI-driven data visualization

Background:

Antimicrobial resistance (AMR) is a pressing global health problem, affecting low- and middle-income countries as well as developed countries. The Antibiotic Resistance Surveillance (ARS) system, based at the Robert Koch Institute, serves as Germany's platform for monitoring antibiotic-resistant pathogens. Laboratories voluntarily contribute data on samples tested and pathogens isolated, as well as resistance test results for a wide range of clinically relevant pathogens on a daily basis. The dataset is complex and high-dimensional, comprising more than 2000 pathogens and 200 antibiotics.

Aim/s:

The primary goal of the project is to develop innovative ML/AI-powered tools to visualize the extensive data collected within ARS. By leveraging machine learning techniques, the project aims to enhance the exploration and understanding of emerging patterns in antimicrobial resistance surveillance data. Specific objectives include identifying changes in resistance patterns among less common pathogens, exploring co-detection patterns of pathogens within specimens, and developing visualization methods that can effectively highlight significant trends while flagging data anomalies.

AI methods:

Using unsupervised learning methods such as anomaly detection and clustering, this project aims to optimize existing surveillance indicators and propose new ones based on the complex AMR data from the ARS system. By using categorical variable coding for multidimensional data and addressing missing data challenges with AI techniques, the project aims to efficiently uncover hidden patterns. Validation will include external data sources and expert feedback to ensure the reliability of the results.

Keywords:

Antimicrobial resistance, Antibiotic Resistance Surveillance, Machine Learning, Artificial Intelligence, Data Visualization, Pathogen Resistance Patterns, Co-detection Patterns, Public Health Surveillance

to the top