Researchers from the University of Geneva Investigate a Graph-based Machine Learning Model to Predict Risks of Inpatient Colonization by Multidrug-Resistant (MDR) Enterobacteriaceae

Machine learning has emerged as a very important tool in healthcare, revolutionizing various aspects of the industry. One of its primary applications lies in diagnostics, where machine learning algorithms analyze vast datasets, including medical images, genetic information, and patient records, to identify patterns and make accurate predictions.

Earlier, machine learning models were used to detect patients susceptible to infection and support Infection Prevention and Control (IPC) programs. Large amounts of medical data regularly gathered in electronic health records (EHRs) were used in these models. While classic machine learning models might show effective results in limited use cases, they fail to generalize to large-scale and longitudinal EHR data.

Researchers at the University of Geneva have made a groundbreaking stride in healthcare technology. They used Graph Neural Networks (GNNs) in healthcare to detect antimicrobial resistance (AMR) and multidrug-resistant (MDR) Enterobacteriaceae colonization.

Enterobacteriaceae are normally found in a healthy human gut, but if they colonize other body parts and cause infections, they can be extremely dangerous to one’s health. Numerous factors contribute to the rise of these pathogens in healthcare environments. 

The researchers modeled the interactions among patients and healthcare workers using a graph structure, where nodes and their interactions describe edges that describe patients. Then, a graph neural network (GNN) model was trained to learn colonization patterns from the patient network enriched with clinical and spatiotemporal features.

Professor Douglas Teodoro, from the University of Geneva, said that the core goal was to model the complex interactions within healthcare environments to predict the spread of healthcare-associated infections (HAIs). Network information about patients and healthcare workers was incorporated into this prediction. He also said that the study’s most crucial message is the potential of analyzing healthcare network interactions to enhance the prediction of HAIs. This method may significantly advance infection prevention and control techniques in healthcare environments.

Teodoro also said that given the method’s data-driven approach, they anticipate its applicability to other pathogens with similar transmission dynamics and in various healthcare settings.

The study includes an image named Graph-Based Prediction of Hospital Infections, showcasing how the team applied Graph Neural Networks to model intricate patterns in transmitting multi-drug resistant Enterobacteriaceae. This research aims to transform how hospitals predict and handle infection risks.

The models were trained and assessed using the Medical Information Mart for Intensive Care (MIMIC-III) dataset and were compared against traditional machine learning baselines. Notably, GNN models performed better predictively than baseline models for the early detection of antimicrobial-sensitive (AMS), AMR, and MDR Enterobacteriaceae.

The researchers tested the model and found that the area under the receiver operating characteristic curve (AUROC) performance was above 88% when patients colonized by vancomycin-resistant Enterococcus were identified using spatiotemporal features. Researchers found the GNN model demonstrates a performance range of 0.91 to 0.96 in the area under the receiver operating characteristic curve (AUROC). This performance is 8% higher than a logistic regression baseline, which scored 0.88. 

Check out the┬áPaper┬áand┬áReference Article.┬áAll credit for this research goes to the researchers of this project. Also,┬ádonÔÇÖt forget to join┬áour 33k+ ML SubReddit,┬á41k+ Facebook Community,┬áDiscord Channel,┬áand┬áEmail Newsletter, where we share the latest AI research news, cool AI projects, and more.

If you like our work, you will love our newsletter..

­čÉŁ Join the Fastest Growing AI Research Newsletter Read by Researchers from Google + NVIDIA + Meta + Stanford + MIT + Microsoft and many others...