Use of AI models to enhance the efficiency and scalability of antibiotic stewardship programs

Nov 21, 2023

SEOUL, South Korea – Enolink announces the publication of collaborative research with Seoul National University Bundang Hospital (SNUBH) and Seoul National University (SNU) featured in the International Journal of Medical Informatics. This study showcases the use of artificial intelligence (AI) and machine learning models to significantly enhance the efficiency and scalability of antibiotic stewardship programs (ASP) at the hospital.

ASP is not just a healthcare protocol but a global mandate, endorsed by the CDC and healthcare leaders worldwide. The process is labor intensive, involving dedicated staff and resources to actively track and manage all patients who are given antimicrobials.  The project aims to streamline this labor-intensive process, reducing the burden on healthcare staff and improving patient outcomes.

Dr. Ju-Yeun Lee, a principal investigator of this research from SNUBH, explains the project's objectives: 

“Our AI model leverages diverse clinical data – diagnoses, drug usage, lab results, microbial culture tests, and vital signs of inpatients on antibiotics. By analyzing this data, the model identifies patients requiring priority intervention in antibiotic management, enhancing the ASP's overall efficacy.”

The collaboration with Enolink has been highly productive, according to Dr. Lee:

“The Enolink team has been instrumental in driving this project. Their systematic approach and regular research discussion have enabled joint decision-making, streamlining the project's progression.”

The Enolink Data Science team, in collaboration with SNUBH’s pharmacy team, developed a sophisticated model prioritizing hospitalized patients for antibiotic stewardship intervention. This model, derived from over 130,000 de-identified daily patient records, predicts the necessity of interventions in antibiotic usage, guiding decisions on stewardship intervention. 

In line with this, the ASP pharmacists who participated in the research stated:

'We need to validate the efficacy of our approach, which entails developing an application that deploys our model. This application will be integrated with the hospital's EMR system or data warehouse system to process real-time patient data, and we’ll apply it in real clinical settings within the hospital.' 

“As a physician who has spent many years involved with antibiotic stewardship efforts for our patients, I am very excited to see Enolink’s successful collaboration to bring an AI/ ML solution to this complex and resource intensive clinical decision making process”, says Dr. Todd Astor, Chief Medical Officer at Enolink.  “We look forward to this continued partnership with SNUBH, and to applying Enolink’s unique data science platform and expertise to tackle many other real world healthcare challenges”.

Enolink and SNUBH plan to jointly launch a pilot project in 2024 based on the recently developed model, aiming to develop practical applications of the model in clinical settings to further improve patient outcomes.

While currently focused on a single center in South Korea, this model promises significant resource optimization for ASPs. Looking ahead, Enolink plans to adopt federated learning, enhancing model training using data from various hospitals. This approach is expected to improve the model's applicability across diverse healthcare settings – a crucial factor given the variations in hospital policies, patient demographics, and regional practices.

Enolink envisions a future where ASP is seamlessly integrated with AI-driven predictions and analytics. The goal is to automate much of the ASP process, reducing time and resources while maintaining high standards of patient care. This ambitious project reflects Enolink’s commitment to pioneering healthcare solutions through technological innovation.