Detecting clinical medication errors with AI enabled wearable cameras.

Area of Science:

• Medical technology
• Artificial intelligence in healthcare
• Patient safety

Background:

• Drug-related errors are a significant source of preventable patient harm in clinical environments.
• Current safety protocols often lack automated checks during critical medication preparation stages.

Purpose of the Study:

• To introduce and evaluate a novel wearable camera system for the automatic detection of potential drug errors.
• To assess the system's capability in identifying and classifying drug labels during medication preparation.

Main Methods:

• Development of a wearable camera system utilizing deep learning algorithms for image recognition.
• Creation of a large-scale, 4K video dataset from head-mounted cameras in real-world operating rooms.
• Evaluation of the system on 418 drug draw events, encompassing routine care and controlled settings.

Main Results:

• The system achieved high accuracy in detecting and classifying drug labels on syringes and vials.
• Demonstrated 99.6% sensitivity and 98.8% specificity in identifying vial swap errors.
• Successful detection of potential errors prior to medication delivery in operational settings.

Conclusions:

• The wearable camera system shows significant potential as an automated secondary check for medication selection.
• This technology offers a real-time opportunity for intervention, thereby preventing medical errors.
• The findings support the integration of AI-powered wearable systems to enhance patient safety in healthcare.