Extracting antibiotic susceptibility from free-text microbiology reports using natural language processing.

Area of Science:

• Infectious Diseases
• Medical Informatics
• Computational Biology

Background:

• Clinical microbiology reports contain valuable antibiotic susceptibility data.
• Manual extraction of this data is time-consuming and inefficient.
• There is a need for automated methods to leverage this information for public health.

Purpose of the Study:

• To develop and evaluate a large language model (LLM) and machine learning approach for extracting antibiotic susceptibility information from free-text clinical microbiology reports.
• To demonstrate the utility of LLM-based systems in infectious disease applications.

Main Methods:

• Utilized natural language processing (NLP) and machine learning algorithms to parse and extract antibiotic susceptibility data from unstructured text.
• Trained models on a diverse dataset of clinical microbiology reports.
• Validated the accuracy and efficiency of the extraction process.

Main Results:

• Successfully extracted antibiotic susceptibility information with high accuracy.
• Demonstrated the potential for LLM-based systems to significantly improve information gathering efficiency.
• Showcased the applicability of the method for real-time outbreak detection and public health surveillance.

Conclusions:

• LLM-based systems offer a powerful tool for automating the extraction of critical antibiotic susceptibility data.
• This approach can enhance infectious disease surveillance, improve outbreak detection, and streamline public health reporting efforts.
• Further development and implementation of LLM in clinical microbiology can advance antimicrobial stewardship and patient care.