Automated medication verification system (AMVS): System based on edge detection and CNN classification drug on...

Area of Science:

• Medical Informatics
• Computer Vision
• Artificial Intelligence

Background:

• Manual medication verification is time-consuming and error-prone, particularly in high-workload hospital environments.
• Existing systems may lack efficiency and accuracy, contributing to medication errors.

Purpose of the Study:

• To develop and evaluate a novel automated medication verification system (AMVS) to improve accuracy and efficiency.
• To reduce medication errors and alleviate the workload of healthcare professionals.

Main Methods:

• Utilized deep learning models for image classification and edge detection to identify and verify medication types.
• Developed a system that classifies multiple medications within a single image without manual labeling.
• Conducted experiments in a controlled, closed environment to minimize optical variations.

Main Results:

• Achieved over 95% accuracy in automated drug recognition and segmentation analysis.
• Demonstrated high accuracy rates: approximately 96% for fewer than ten drug types and 93% for ten drug types.
• Successfully developed a fully automated drug recognition system.

Conclusions:

• The novel AMVS effectively automates medication verification with high accuracy.
• The system shows significant potential in assisting nursing staff and reducing medication errors.
• This technology can enhance efficiency and safety in hospital medication management.