Beyond the human eye: Artificial intelligence revolutionizing plasma quality control

Area of Science:

• Transfusion Medicine
• Medical Technology
• Artificial Intelligence in Healthcare

Background:

• Current plasma quality control (QC) relies on subjective visual inspection, leading to variability due to technologist experience and lighting.
• This subjectivity can compromise the safety and efficacy of transfused plasma products.
• There is a need for standardized, objective methods for plasma QC.

Purpose of the Study:

• To develop and evaluate an automated system for detecting plasma color and turbidity abnormalities.
• To assess the accuracy and reliability of this automated system in a transfusion setting.
• To compare the automated system's performance against traditional manual inspection methods.

Main Methods:

• An automated system using a sorting machine and deep learning model was developed for plasma QC.
• The system captured images under controlled lighting for automated quality assessment.
• Performance metrics including accuracy, sensitivity, and specificity were evaluated across multiple testing phases.

Main Results:

• The automated system demonstrated high accuracy, reaching 94.7% in Phase 2.
• Sensitivity was 100%, ensuring no abnormal plasma units were missed.
• Specificity was 90.9%, indicating reliable identification of normal units.

Conclusions:

• The automated QC system provides a reliable and objective alternative to manual plasma inspection.
• It minimizes errors, reduces wastage, and enhances blood bank efficiency.
• The system's high sensitivity and integration capabilities support improved patient safety in transfusion medicine.