Applications of Machine Learning on Electronic Health Record Data to Combat Antibiotic Resistance

Area of Science:

• Biomedical Informatics
• Computational Biology
• Clinical Research

Background:

• Growing interest in artificial intelligence (AI) and machine learning (ML) for clinical applications.
• Advancements in computing and ML frameworks facilitate predictive model training with electronic health record (EHR) data.

Purpose of the Study:

• To provide a primer on ML techniques applicable to EHR data.
• To explore the application of ML in addressing antimicrobial resistance (AMR).

Main Methods:

• Review of ML approaches for EHR data challenges.
• Case studies on using EHR data for ML models in AMR: predicting pathogen carriage/infection, optimizing empiric therapy, and supporting antimicrobial stewardship.

Main Results:

• ML models can be constructed using EHR data for various AMR-related tasks.
• Demonstrated potential of ML in promoting appropriate antimicrobial use.

Conclusions:

• ML holds significant promise for enhancing antimicrobial stewardship and combating resistance.
• Clinical deployment of ML models faces limitations due to practical and implementation barriers.